# Tana > Tana is an agentic meeting platform. Work gets done in the meeting, not after. Agents create documents, file issues, and update your context graph while you talk. And each meeting makes future meetings smarter. This file contains the full documentation for **Tana**, an agentic meeting platform where AI agents do real work during meetings. **Tana Outliner** is a separate, earlier product for knowledge management through structured outlining. It lives at https://outliner.tana.inc and has its own documentation there. These are distinct products that share a brand name. If you are looking for help with Tana Outliner, do not use this documentation. ## Tana ### Features #### Agents URL: https://tana.inc/learn/features/agents Agents are custom AI assistants you build by describing what you need and adjust the same way, by talking to them. - [Build an agent](#creating-an-agent) by describing what you need; the AI scaffolds the prompt, capabilities, and settings - [Capabilities](#agent-capabilities) bundle the tools an agent needs with the instructions for using them well - Run agents in [chat](#chatting-with-an-agent), by [voice](#voice-mode), or on a [schedule](#schedules) so they run while you sleep - Each [space](/learn/features/spaces) can set a default agent that picks up new chats automatically - [System agents](#system-agents) are built in and read-only; you build the rest ## What an agent is An agent is an AI assistant you configure for a specific job: prepping you for meetings, triaging bugs, drafting weekly updates, or anything else you would otherwise repeat in a fresh chat. Each agent has its own capabilities, prompt, settings, and optional schedules, and lives as a doc you can open like any other. The default Tana assistant is one such agent, built in to every org. Custom agents stand alongside it. ## Creating an agent Click the **pencil button** at the top of the sidebar to start a new agent, then describe what you want. Agents created from here are global, available to the whole organization, and show up in the sidebar. You can also ask the AI chat to create an agent: "I want an agent that preps me for every meeting by pulling context on the people I'm meeting with." Tana builds the agent, including name, prompt, capabilities, and a starting set of context fields. Adjust it the same way you created it: keep talking to it, in [chat](#chatting-with-an-agent) or by [voice](#voice-mode). The agent edits its own configuration as you tell it what to change. Custom agents live in [spaces](/learn/features/spaces), so they pick up that space's context and members. Agents owned by a space appear as a leaf row under that space in the sidebar tree, revealed via the chevron just like a sub-space. Global agents (not scoped to a space) show in a **Custom agents** section in the sidebar. Agents in the sidebar are draggable, so you can re-parent an agent to another space or drop it into an editor as a link. ## Customizing an agent Open the agent and click **Customize** next to the title. The panel expands with sections for: - **Capabilities:** What the agent can read, do, and connect to, including [integrations](#agent-capabilities) like GitHub, Slack, Linear, Jira, and HubSpot. - **Prompt:** The instructions the agent follows, written in a rich text editor with full formatting. An **Improve Your Agent** helper opens a chat that can read and rewrite the prompt with you. To include Tana docs, chats, or meetings as context, reference them in the prompt with `@` mention search. - **Schedules:** Days and times for the agent to run automatically. See [Schedules](#schedules). The agent doc also lists recent outputs and chats so you can see what the agent has been doing. ### Agent capabilities The **Capabilities** section of the Customize panel controls what the agent can do, organized into two groups of skills. **Capability skills** bundle a coherent set of tools with the instructions for using them well: - **Reading capabilities:** Searching, reading, and gathering context from Tana. - **Create capabilities:** Creating new documents and items. - **Update capabilities:** Editing existing documents, fields, and structure. Toggle a capability on to grant the agent its underlying tools plus Tana's guidance for using them. As Tana refines a capability over time (adding tools, sharpening instructions), every agent that has it picks up the improvement automatically. New agents are seeded with the **Reading** capability so they are useful out of the box. **Integration skills** appear as a dedicated section with one-click toggles for each connected service: GitHub, Slack, Linear, Jira, and HubSpot. Each row shows the live connection status. If an integration is not connected, the row shows a **Not connected** button that deep-links straight to its setting in [Integrations](/learn/features/integrations). For the rare case where you need a single tool that no capability or integration covers, an **Advanced** picker lets you enable individual tools directly. Capability skills are the preferred path because they bundle tools with their usage instructions and stay current as Tana improves them. Agents that were configured before the switch to capabilities get an upgrade button on the agent doc that converts their old tool-bundle setup to the matching capabilities. ### Default agent per space Every [space](/learn/features/spaces) can set a default agent. Any new chat started from that space's home picks up the default agent unless you choose a different one explicitly. To create an agent inside a space, navigate to the space and click the **+** button in the top right corner, then select **Agent**. You can also create the agent in chat and ask the AI to move it to the space. ### Auto-resume after proposals Agents can be configured to continue the conversation automatically once you have accepted or dismissed the last pending [proposal](/learn/features/proposals), so multi-step flows do not stall waiting for a nudge. ## Chatting with an agent Open the agent and start typing in its chat panel. It uses the same composer as everywhere else in Tana ([@mentions](/learn/features/chat#mentions-and-skills), [skills](/learn/features/skills) with `/`, attachments, voice input). The agent uses its configured capabilities, prompt, and context fields on every message. You can also reference an agent from any chat with `@`, the same way you reference a document, or pick which agent answers from the agent picker in the compose bar. The agent is timezone-aware: when you ask about "today" or "yesterday", it uses your own calendar day, even if the chat spans midnight or a teammate is in a different time zone. ## Voice mode Agents can run in voice mode, letting you have a two-way live voice chat with them. Voice happens in a meeting. Open or create one and start the agent from the voice control at the head of the meeting digest, or say the wake word in the call. The voice your organization uses is set once for everyone rather than per agent. For what the voice agent can do, including following the meeting, running work in the background while you keep talking, and presenting a document on request, see [Meetings, Voice agent](/learn/features/meetings#voice-agent). ## Schedules Put an agent on a schedule so it runs without you. In the Customize panel's **Schedules** section, pick days of the week and a time of day. Each schedule has a **Timezone** (a searchable picker that defaults to your own zone), so a daily brief set to 09:00 in Oslo always lands on the right Oslo calendar day, and the agent reasons about "today" in that zone. Each schedule can carry an optional **Additional prompt** that varies what the agent does on different runs; leave it blank and the run starts with a standard message. For example, a daily morning agent might pull a different focus area depending on the day. You can schedule any agent you can open, including ones shared with you. Schedules are private: you see and manage only the ones you created. The agent's activity panel shows enabled schedules with the routine in its configured zone and the zone labelled (for example, "Weekdays at 09:05 (Europe/London)"). The **Next run** and **Last run** labels stay in your own local zone. ## System agents Some agents are built in to Tana and managed centrally. They show a **System agent** badge instead of the Customize button and cannot be edited. The default Tana assistant is the canonical example. ## Deleting an agent Open an agent's doc and choose **Delete** from the overflow menu. ## Reporting issues If an agent produces an incorrect or unhelpful response, right-click the message and select **Report Bad Chat**. The full conversation, plus the agent's current configuration, is attached automatically for analysis. --- #### AI chat URL: https://tana.inc/learn/features/chat Chat with AI on your own, or bring AI into a conversation with teammates by @mentioning it. - AI can read, search, create, update, and delete content across Tana - Reference any document with [@mentions](#mentions-and-skills) and invoke [skills](/learn/features/skills) with / - Every AI change is a [proposal](/learn/features/proposals) you review before it applies - Connected to [GitHub, Slack, Linear, Jira, HubSpot, and coding tools](#connected-tools) for cross-tool workflows - [A search asked in chat](#search-results-in-chat) comes back as a result card you can save - [Voice input](#voice-input) with real-time dictation in the chat input - Switch between [models](#choosing-a-model) from OpenAI, Anthropic, and Google ## Solo chat or team chat Every chat in Tana is either a solo conversation with the AI or a conversation with teammates. The two work the same way, with one difference: - **Solo chat:** You and the AI. The AI responds to every message. - **Team chat:** You, teammates, and optionally the AI. The AI stays quiet by default. Mention @Tana when you want a response, so human conversation is not interrupted. In both cases, the AI has the same capabilities: reading your content, creating and updating documents, invoking [skills](/learn/features/skills), and connecting to your [integrations](/learn/features/integrations) and [custom MCP servers](/learn/features/integrations#custom-mcp-servers). ## Starting a chat There are several ways to start a conversation with AI in Tana: - **Calendar:** The chat bar on the [Calendar](/learn/features/calendar#asking-about-what-you-see) takes what you are looking at as its context, the visible range or the meeting you have open. - **Space home:** The "Ask or start a chat..." bar at the bottom of every space home. Type a message and press enter to create a new chat scoped to that space. - **Meetings:** Every meeting has two chat surfaces. The private [Ask or Capture](/learn/features/meetings#chat-in-the-call) pill in the bottom bar is yours alone and the AI auto-responds to every message. The shared [Chat With Everyone](/learn/features/meetings#chat-in-the-call), opened from the call toolbar, is visible to all attendees and stays quiet until you mention @Tana. - **Quick chat:** Press on the desktop app to open a new chat from anywhere. Press to open it with voice dictation active. - **Sidebar:** Click "New Chat" in the sidebar. - **Create menu:** Select "Chat" from the create menu. - **Documents:** Many document types have a contextual chat panel that scopes the conversation to the open document. Recent chats appear in the sidebar sorted by last activity, with unread indicators for new messages. ## Composing a message ### Mentions and skills Reference documents and invoke AI capabilities directly from the message input: - Pick which **agent** answers from the picker in the compose bar. Chats run against the built-in Tana agent unless you or the space set another. See [Agents](/learn/features/agents). - Type **@** to search and reference any document in Tana. The AI receives the referenced document as context. Mentioning @Tana triggers an AI response in chats with other people. - Type **/** to search and attach a [skill](/learn/features/skills). Picking one runs it straight away, using whatever you have typed as its input, or its own defaults when you leave the message empty. If what you typed is an instruction about the skill itself, the AI answers that instead of running it. The same `/` also accesses emoji and formatting commands. - Type to open the emoji picker (about 250 common emojis searchable by shortcode). Press again to close the picker without inserting anything. ### Attachments Add images, audio, video, PDFs, and other files by dragging them into the chat, pasting, or clicking the attach button. Each attachment shows upload progress while it is sending, and Send waits until uploads finish so nothing is silently dropped. Failed uploads can be removed and retried. Images get AI-generated descriptions automatically, and the AI reads PDFs, audio, and video so it can summarize them, answer questions about their content, and embed media inline in its replies. ### Voice input The chat toolbar has two voice modes: - **Dictation:** Click the microphone button to speak your message. Words appear in the input as you speak. Supports English, Norwegian, and many other languages. - **Audio recording:** Click the voice button to record an audio clip, preview it, and send as an attachment. The audio is automatically transcribed. ## What AI can do ### In Tana The AI has full access to your content (within your permissions). It can: - **Read:** Documents, types, fields, meeting transcripts and participants, screen capture screenshots, calendar events, skill definitions, previous chats, tables, attached PDFs, and uploaded audio and video - **Search:** Keyword search, semantic search, relationships between documents - **Create:** Docs, tasks, typed entries, types with fields, skills, artifacts, AI-generated images, and calendar events - **Update:** Titles, content, fields, calendar events, space configuration, table data, and type definitions - **Delete:** Any document - **Link:** Build a shareable web link to a document or a meeting, for pasting into another tool or sending to a teammate Ask for a link when you want one that works outside the chat, in an email, a CRM record, or an issue tracker. The link opens Tana in the browser and the person opening it has to be signed in to your organization, so it is not a public link. Inside the chat you do not need one: documents the AI mentions are already clickable. When the AI creates a calendar event it needs a title and start time; end time, timezone, location, description, and participants are optional. Passing participants restricts the event to that list with you as admin. It can propose changes to an existing event's title, description, time, timezone, location, mode, and existing tagline or summary. For a synced event, title and time changes require two-way calendar sync and the connected calendar's organizer. Attendees and meeting links remain calendar-sourced. Use `pinItem` to pin documents to an event. Updates and deletes and creates all appear as [proposals](/learn/features/proposals) for your review. AI never changes anything without your approval. ### Connected tools When you connect integrations, the AI can act on them from chat: - **GitHub:** Search and create issues, review and merge pull requests, read code and files, comment on issues and PRs - **Slack:** Send messages to channels or people, list channels and users - **Outlook:** Search and read your mail, draft a reply, and send a message once you accept the action that proposes it - **Linear:** Search, create, and update issues - **Jira:** Search, create, and update issues, with meeting screenshots embedded as visual context - **HubSpot:** Search contacts, companies, and deals, update properties and lifecycle stages, create deals, and log notes to CRM records - **Custom MCP servers:** Use the tools from any [remote MCP server you connect](/learn/features/integrations#custom-mcp-servers), surfaced through the linked skill for that server - **Coding tools:** Hand off the current document or chat as a self-contained prompt to Claude Code, Cursor, Codex, GitHub Copilot, or Lovable. Triggered by phrasings like "create a PR" or "send this to Cursor"; when more than one tool is enabled, you pick. Configure integrations in Settings. See [Integrations](/learn/features/integrations). ### Managing access The AI can also propose changes to a document's access: restricting it to specific participants with editor or admin roles, opening it back up to everyone in its space or event, and toggling the Guest Link (anyone with the link can view). Access changes go through the same proposal approval flow as the rest of the AI's edits, and the proposal preview shows a clear access callout before you accept. ### Default content language Every organization picks a default content language in Settings, Organization, Language, from 33 languages. The AI writes its replies, generated artifacts, outcomes, summaries, and document titles in that language no matter which language the conversation is happening in. Meeting transcription stays multilingual (auto-detected per speaker) so a mixed-language call still transcribes everyone correctly. The Tana interface itself is English. ### Usage limits Your monthly AI credits are the one limit you can manage. When you have used 80 percent of them, an amber callout appears at the top of chat ("You've used N% of your monthly AI credits") and a warning lands in [Work](/learn/features/interface#work). When your credits run out, chat shows the same style of callout instead of a reply, with a "Resets in..." countdown and, where billing is available, a **Top up** button that opens Billing settings. Separate safety limits also protect against runaway usage (for example, a looping automation); if one trips, the callout shows a reset countdown but no Top up button, because a purchase cannot clear a safety limit. A reply cut off mid-stream is marked with a compact "Usage limit reached" badge. ### Web search and image generation Models that support web search can look up current information when your question goes beyond Tana. Results include citation links to sources. Web search availability varies by model. The AI can also generate images from text descriptions. Generated images become image documents owned by the chat they were created in, so an image made in an "Only me" chat stays private. When a reply includes a Mermaid diagram, Tana renders it inline once the diagram is complete, so you never see a half-drawn render. A malformed diagram shows a compact notice instead, and the AI corrects it on its next turn. ## Search results in chat Ask the AI something that is really a search, "the open decisions from this quarter", "feedback filed per customer this month", and the answer comes back as a **result card** rather than prose. Collapsed it shows a type badge and a count. Opened it is a table of the matching documents, with the attribute columns that type carries. **Show more** pages through larger sets without reloading the card. The count reads how many documents matched, not how many are drawn. Where the table shows fewer, a note says so, so a total you take from the rows is a floor rather than the answer. - **Save it.** The card carries its own **Save search** action, so a search the AI ran becomes a [saved search](/learn/features/interface#saved-searches) document you can pin, exactly like one you built yourself. - **Searches attached to a type.** A [type](/learn/features/types) can carry the searches worth running on it, listed on its panel. The AI can run one of those, or attach one to the conversation. - **An existing saved search wins.** Where a saved search in your organization already answers the question, the AI shows you that one instead of composing a new query. ## Reviewing changes Every AI-generated document, edit, or deletion appears as a proposal you review before it takes effect. Proposals show up as inline cards on the message that produced them, so you can see what the AI did and decide on it in place. Each pending proposal shows a **Lives in** callout for where the new document will live, alongside an access callout for who will see it, so the consequences are visible before you accept. Accept individually, reject, or iterate by chatting with the AI to refine before accepting. When the AI assigns a task to you while acting on your behalf, you do not get a notification for your own work. Assigning to someone else still notifies them, attributed to you. Writing to a connected service works the same way, as an **action** you approve before anything leaves Tana. See [Proposals and actions](/learn/features/proposals) for the full review workflow. ## Choosing a model Tana supports models from OpenAI, Anthropic, and Google. Switch models at any time during a conversation. The change takes effect on the next message. Models are grouped by cost: - **Cheap:** Smaller, faster models for simple tasks - **Moderate:** Balanced cost and capability - **Expensive:** Most capable models for complex reasoning Some models support **reasoning** with configurable effort levels (low, medium, high). Some support **web search**. Both capabilities are indicated in the model selector. ## Clarifying questions The AI may ask clarifying questions before proceeding with a task. When this happens, the chat input is replaced with the question and a set of options. Each card also includes a **Something else** free-text fallback at the bottom so you can answer in your own words instead of being forced into one of the suggested choices. Pick an option, type a custom answer, or skip the question. ## Side panels The chat toolbar opens side panels for: - **Outcomes:** Pending proposals and accepted outcomes from this conversation - **In common:** Documents shared between chat participants. Only visible when the chat has multiple participants. ## Agents Beyond the default Tana assistant, you can build custom agents with their own capabilities, prompt, voice, and schedules. Click the pencil button at the top of the sidebar, describe what you want, and Tana scaffolds the agent for you. You can also ask the AI to find one by what it does, for example "find the agent that does customer research", and it searches your own agents by name and purpose. See [Agents](/learn/features/agents) for the full guide. ## Reporting issues If the AI produces an incorrect or unhelpful response, right-click the message and select "Report Bad Chat." Describe what went wrong and the full conversation is attached automatically for analysis. When an AI run fails outright, the chat displays a red callout with a one-line, actionable reason: the model rejected the request (quoting its reason, for example an attached image being too large), the response timed out, or a generic "Something went wrong" fallback. Cancelled and "Usage limit reached" keep their pill style. Transient problems are retried automatically, falling back to another model rather than surfacing as an error. --- #### Apps URL: https://tana.inc/learn/features/apps Tana runs in the browser and as a desktop app for Mac, Windows, and Linux, all syncing in real time. - [Desktop app](#desktop) for Mac, Windows, and Linux: tabs, split view, and [external meeting capture](/learn/features/meetings#capture-an-external-call) - [System tray](#system-tray) shows today's agenda for every signed-in org - [Web app](#web): works in any browser, installable as a PWA - All platforms [sync in real time](#sync) - Deep links (`tana-app://`) for opening specific documents from other tools - A Tana link in your browser can [hand off to the desktop app](#opening-a-link-in-the-desktop-app), and a new tab signs in to that link's organization ## Desktop The Tana desktop app is available for Mac, Windows, and Linux, and provides features beyond the web experience. Download it from the [download page](/download), which detects your operating system, or from the Download button in the top navigation. The Windows download is a small installer that fetches the app while it installs, so the file you download is a couple of megabytes rather than the whole app. ### Tabs Open multiple documents in separate tabs within one window. The tab bar appears at the top when you open a second tab. Each tab maintains its own navigation history and split view state. A new tab signs in to the organization of the link that opened it, not whichever organization you used last, so following a link into another workspace does not move your current one. Use to open a new tab, to close the current one, and or to cycle between them. through the eighth tab jumps straight to that tab, and jumps to the last one. on any link opens it in a new tab. ### Split view Work with two documents side by side. Use on any link to open it in the other pane. Drag the divider to adjust the split ratio. Each pane can be closed independently. Use on a link to open it in a new window instead. ### Find in page Use to open a find bar that searches the current page. Enter jumps to the next match, Shift and Enter to the previous, and Escape closes it. ### Profiles The **Profiles** menu in the application menu bar lists your profiles and **New Profile...**, which opens a window with its own isolated sign-in. Use profiles to stay signed in to multiple accounts or organizations side by side, each in its own window. ### External meeting capture The desktop app can detect active Zoom, Teams, and Google Meet calls and capture their audio and screen content for transcription and outcome extraction. See [External meeting capture](/learn/features/meetings#capture-an-external-call) for the full workflow. ### Launch at login **Launch at Login** in the app menu on macOS, or **Launch at Startup** in the File menu on Windows, opens Tana when you log in. It is on by default; untick it to stop. ### System tray The desktop app installs a menu bar icon that shows today's agenda across every organization you are signed in to. Events are listed in time order. When you are in more than one org, each event is labelled with the org it belongs to so it is clear which org it is from. The menu bar title shows the meeting that is happening right now, or the next one coming up if nothing is in progress. Click any event to jump to it. ### Deep links The desktop app registers the `tana-app://` URL scheme. Links in this format open the corresponding document directly in the app, making it easy to link to Tana content from other tools. ### Opening a link in the desktop app Click an organization-scoped Tana link in a browser with the desktop app installed and Tana offers to hand it off before asking you to sign in. The **Open in Tana** chooser offers **Open always**, which opens it now and saves the preference, **Open once** for this time only, and **Continue in browser**, which is never saved. With the desktop preference saved, later links show a brief "Opening in the desktop app" screen with an **Open in browser instead** escape and an **Always ask where to open** button to clear the preference. You can also reset it from the **Tana links** row in Settings, Preferences. Meeting links skip this entirely, so a guest joining a call in the browser is never sent to the app. ## Web Tana works in any modern browser at [home.tana.inc](https://home.tana.inc). The web app can be installed as a PWA (Progressive Web App) for a more app-like experience. The web app has the same core workspace and meeting features as the desktop app. Desktop features, including tabs, split view, profiles, external meeting capture, the menu bar agenda, and `tana-app://` links, are not available in the browser. ## Sync Tana syncs in real time across all your devices. The sync indicator in the sidebar shows connection status. ## Updates When a new version is ready, Tana shows an update banner. How soon it appears depends on how long your session has been open, and after a long idle stretch Tana asks you to reload with a "Reload required" prompt. Neither the banner nor the prompt interrupts you while you are in a call. On macOS, automatic updates need a macOS version the current app still supports. On an older macOS, Tana keeps the version you have rather than installing an update that would not run. --- #### Artifacts URL: https://tana.inc/learn/features/artifacts Artifacts are AI-generated visual documents that turn conversations into presentation-ready material. - Three types: [storyboards](#storyboard), [customer journey maps](#customer-journey), and [slide decks](#slides) - Generated from [meeting transcripts](/learn/features/meetings#outcomes) or [chat conversations](/learn/features/chat) - Fully [editable](#editing-artifacts): click any text to edit inline, drag to reorder, AI to iterate - Created as [proposals](/learn/features/proposals) you review before accepting - Shareable like any document via [access controls](/learn/features/spaces#access-modes) ## What artifacts are Artifacts are structured visual documents generated by AI from your conversations. Unlike docs (which are freeform text), artifacts have a specific layout and visual design tailored to their type. They turn meeting discussions into presentation-ready materials. ## Creating artifacts Generate artifacts from two places: - **Meeting outcomes:** Click "Generate artifact" in the outcomes panel during or after a [meeting](/learn/features/meetings). Choose the artifact type and AI generates it from the transcript and screen captures. - **Chat:** Ask the AI in any [chat](/learn/features/chat) to create an artifact. The AI uses the conversation context to generate it. Artifacts are created as [proposals](/learn/features/proposals) that you review before they become visible in Tana. ## Artifact types ### Storyboard A visual narrative with scene cards, designed for pitch decks and presentations. Each scene has a title, description, visual notes, and an optional AI-generated image. Scenes are laid out in a scrollable grid. You can drag scenes to reorder them, move scenes to a "Probably not relevant" section to set them aside without deleting, or remove scenes entirely. ### Customer journey A before-and-after comparison showing workflow improvements. The "before" side shows pain points and quotes, the "after" side shows the improved workflow with success criteria. Toggle between before and after views to compare. ### Slides A full presentation deck with a hero slide, takeaway cards, themed sections, action items, and a closing slide. The deck includes its own navigation controls, progress bar, and keyboard shortcuts for presenting. Sections support multiple component types including callouts, checklists, numbered grids, and screenshot galleries. ## Editing artifacts All artifact text is editable inline. Click any title, description, or text element to edit it directly. Changes are saved automatically. For larger changes, open the artifact in [chat](/learn/features/chat) ("Open in chat" in the toolbar) and describe what you want changed. The AI can update specific items, add new content, remove sections, reorder elements, and regenerate images. When reviewing an artifact as a [proposal](/learn/features/proposals), use the iteration chat to refine it before accepting. Each round of changes creates a new version you can compare. ## Sharing Artifacts support the same [access controls](/learn/features/spaces#access-modes) as any document. Share with specific people or make available to everyone in the space. Link sharing also works for storyboards, customer journey maps, and slide decks. Generated images and screen-share screenshots resolve to signed URLs in the public viewer so all imagery renders, and slide decks with full-bleed layouts use the entire viewport. Artifacts opened from a public link are fully read-only: text is non-editable, drag handles and drop zones for storyboard scene reorder are hidden, and delete buttons are removed. Open the same artifact inside the app and inline editing, drag-and-drop reorder, and delete actions are all available. --- #### Calendar URL: https://tana.inc/learn/features/calendar The Calendar shows your meetings over time as day, week, and month grids, with the details of any meeting beside the grid and a chat bar scoped to what you are looking at. - [Day, week, and month](#day-week-and-month) grids of your [meetings](/learn/features/meetings), opening on the week - [Click a meeting](#a-meeting-beside-the-grid) for its details beside the grid, editable before it has happened - [A wrapped-up meeting](#a-meeting-beside-the-grid) shows its summary, outcomes, and proposals still for review - [Drag on the grid](#drafting-a-meeting) to draft a meeting, or [step between dates](#navigating-between-dates) by day, week, or month - [Ask about what you are looking at](#asking-about-what-you-see): the chat bar is scoped to the visible range or the open meeting - An [Attendees filter](#day-week-and-month) narrows the grid to meetings involving particular people ## Day, week, and month The Calendar is your meetings over time, as **Day**, **Week**, or **Month**. It opens on Week, and the sidebar's Calendar row stays lit across all three. An **Attendees** filter narrows the grid to meetings involving particular people. Opening a date from anywhere else in Tana, a date mention or an item pinned to a date, opens that day on the Calendar's Day view. The Calendar has no search of its own and no space scope: a search spans every date, which is the opposite of what a dated view is for. To reach a meeting outside the visible range, use [Search](/learn/features/interface#search) or **More**, **Meetings**, both of which keep the date and attendee filters. The Calendar shows your own calendar events and organization-level events synced by other members. See [Calendar sync](/learn/features/integrations#calendar-sync) for connecting Google Calendar or Outlook. ## A meeting beside the grid Click a meeting in the grid and its details open beside it. - **A meeting that has not happened yet** opens ready to edit: title, time, and attendees. - **A meeting that has wrapped up** shows what came of it: the summary artwork, how long it ran, its tagline, the accepted [outcomes](/learn/features/meetings#outcomes), the [proposals](/learn/features/proposals) still **For Review**, and anything pinned to it. Press a proposal row to review it there. ## Navigating between dates The Day, Week, and Month switch sets how much you see at once, and stepping forward or back moves by that much. A control returns you to today whenever you are looking at another date. With no explicit date, the Calendar tracks the clock: meetings roll from in progress to past as they end, and the date advances across midnight without a reload. ## Joining from the grid A meeting within 10 minutes of its start, and before its end, carries a **Join** button. Hovering it opens a device-check preview with the camera feed and mic, camera, and settings toggles; clicking joins the call directly. An external meeting carries the provider's join link instead. See [External meetings](/learn/features/meetings#external-meetings). ## Drafting a meeting Drag across the grid to sketch a span, then name the draft and confirm it, or discard it. For the full meeting card, with the people picker and the choice between starting now and scheduling, see [Creating a meeting](/learn/features/meetings#creating-a-meeting). ## Asking about what you see A chat bar on the Calendar takes what you are looking at as its context: the visible range, or the meeting you have open. So "what did we decide?" is scoped to that meeting, and a question about the week is scoped to that week. See [AI chat](/learn/features/chat) for what you can do in a conversation. ## Meeting cards in the grid Each meeting shows its time, title, and attendees, and either a **Join** button for Tana calls or the external provider. When someone is already in a call, the card shows who is in it. - **Click** opens the meeting. - **Drag** an event onto the sidebar to pin it, or into another item to reference it. - **Right-click** opens the standard context menu (Pin to space, Pin to date, Copy URL, Open in new tab), plus **Open in side panel** and **Open in new window** on the desktop app. - opens it in a new tab. On desktop, opens a new window and a side panel. This works on most items, not just meetings. ## No calendar connected With no synced calendar, the Calendar shows a card with Google and Outlook options in place of your events. Dismiss it from its ⋯ menu and it collapses to a **Connect Calendar** button so you can still sync later. See [Calendar sync](/learn/features/integrations#calendar-sync). --- #### Canvas URL: https://tana.inc/learn/features/canvas Canvas is an infinite whiteboard in Tana for visual thinking. - Infinite whiteboard with drawing tools, shapes, arrows, and text - [Embed documents](#embedding-documents) from Tana directly on the canvas - [Real-time collaboration](#collaboration) with live cursors and follow mode - Full editing toolkit: [align, distribute, flip, group](#editing), undo/redo - Every canvas supports [backlinks](/learn/features/docs#backlinks) and [sharing](/learn/features/spaces#access-modes) ## Creating a canvas Select "Canvas" from the [create menu](/learn/features/interface#create-menu). A canvas opens as a full-screen whiteboard with an editable title and drawing tools. ## Drawing tools The toolbar at the bottom of the canvas provides standard drawing tools: select, draw, arrow, text, and shapes. These are the default tldraw tools for freeform visual work. ## Embedding documents Click the embed tool in the toolbar to place a document from Tana onto the canvas. A picker appears showing documents in the current space. Select one to embed it as a card on the canvas. Double-click an embedded document to view its content. You can also create new docs directly from the canvas. When you embed an image, it converts to a native canvas image for better handling. ## Editing The menu bar at the top provides: - **Edit:** Undo, redo, cut, copy, paste, duplicate, delete, select all - **View:** Theme (light, dark, system), zoom controls (in, out, fit, 100%) - **Selection:** Align (left, center, right, top, middle, bottom), distribute (horizontal, vertical), flip (horizontal, vertical), group and ungroup ## Collaboration Multiple people can work on the same canvas simultaneously. Each person's cursor is visible in real time with their name and a color. Use the collaborator panel to follow someone else's view as they navigate the canvas. ## Side panels - **Backlinks:** Documents that reference this canvas - **In common:** Documents shared between canvas participants ## Send to Canvases have the same **Send to** menu as any other document in the panel toolbar. Choose **Start New Chat** to open an AI [conversation](/learn/features/chat) with the canvas attached as context, send it to a teammate or agent, run a skill, copy a link, or download it. See [Send to](/learn/features/docs#send-to) for the full menu. --- #### Docs URL: https://tana.inc/learn/features/docs Docs are the default kind of item in Tana for writing, thinking, and capturing context and information. - Rich text editor with [slash commands](#slash-commands) for headings, lists, code blocks, math, diagrams, and more - Assign a [type](/learn/features/types) to add structured fields, or use as a freeform doc - [Mention](#mentions) docs, people, and items with rich, interactive pills - [Backlinks](#backlinks) show every item that references this doc - [Send to](#send-to) routes a doc to a chat, a skill, a meeting, a copied link, or Markdown export ## Creating a doc Select "Doc" from the [create menu](/learn/features/interface#create-menu), or from the "+" button on any space home or date. A Doc is the default item kind for freeform writing, and the create menu, types table, space home buttons, slash command, and AI tools all use the label **Doc**. The slash-menu aliases "note" and "document" still match the same option, so existing muscle memory keeps working. When you create a typed entry inline (from the kanban board's "+" buttons or the slash command), the **Inline document popover** protects in-flight work: while the new doc is still in proposal state, clicking outside or pressing Escape no longer discards what you typed. Only the **Create** button (to keep it) and the **Discard** trash icon (to drop it) close the popover. Editing an existing doc is unchanged: outside-click and Escape still dismiss the popover. ## The editor A doc opens as a full-screen editor with an editable title at the top and a rich text area below. The title and the body behave as one piece of text for the arrow keys: pressing down on the title's last line moves into the body, up on the body's first line returns to the title, and the caret keeps its place across the line. The editor supports real-time collaborative editing: multiple people can write in the same doc at the same time. ### Formatting Select text to access inline formatting: **bold**, _italic_, ~~strikethrough~~, and `inline code`. Selecting with a mouse or trackpad opens a formatting toolbar above the selection. **Turn into** changes the block to text, a heading, a bulleted, numbered or to-do list, a blockquote, or a code block, and the inline formats sit beside it, each shown pressed when the selection already carries it. On touch the toolbar stays out of the way, because the native selection handles occupy the same space. Everything on it is also in the slash menu, , and the keyboard shortcuts. ### Slash commands Type `/` anywhere in the editor to insert: - **Heading 1, 2, 3:** Section headings at different levels - **Bullet list:** Unordered list - **Numbered list:** Ordered list - **Code block:** Syntax-highlighted code - **Quote:** Blockquote - **Divider:** Horizontal rule - **Math block:** LaTeX equation - **Audio:** an audio file, which plays inline. You can also drop or paste one straight into the editor, the same as an image or video - **Mermaid diagram:** A diagram written in Mermaid syntax In the title of an untyped doc, type `/` to give it a [type](/learn/features/types). Changing or removing a type it already has is done from the panel's **⋮** menu. A Mermaid block renders the diagram with a "Preview" / "Source" toggle (Preview by default); click the diagram to open it in a pan and zoom view. You can also turn any existing code block into a diagram by changing its language to Mermaid. Flowcharts and the common graph types are supported. ### Emoji picker Type in the editor or the chat composer, then at least two letters, to search emoji by name (for example `:thu` finds 👍). Pick one to insert it. Pressing again closes the search without inserting anything. ### Images and attachments Drag and drop, paste, or use the attach button to add images, audio, and video files to a doc. Images get AI-generated descriptions automatically. Video and audio render inline as native HTML5 players in the editor, the read-only viewer, the diff preview, and the compact editor, so attached media plays in place rather than as a separate download. Opening a video or audio item on its own shows it in a dedicated player panel with the standard toolbar, with upload-in-progress and failed states surfaced in the player area. ### Mentions Type `@` to mention another document, a person, or any item. You can also paste a link to another Tana item and it drops in an inline mention at your cursor, while links to images and videos render the media inline. Each mention is a pill with a leading symbol that matches what it points to: a task's state, a person's avatar, an image thumbnail, a type badge for a typed item, or the document-kind icon for everything else (Doc, Canvas, Meeting, Chat, Skill, Agent, Space, and audio, video, or other assets). Long titles truncate so the pill stays one bounded shape, including inside table cells and headings. Move between mentions with the arrow keys, and open one by pressing Enter or clicking: a Doc opens in a peek side panel, other kinds navigate to the item. Modifier-clicks open the mention in a new tab, window, or split, the same as any other link. You can also drag a mention to move it within the doc. ## Types and fields Any item can be assigned a [type](/learn/features/types) to add structured fields. For example, assign a "Bug" type to add fields like priority, assignee, and status. Set the type from the overflow menu ("Set type") or via slash command in the title. Fields render inline on the document: a member field as an avatar, a link field as the referenced title, and an [Options field](/learn/features/types#options-fields) as a chip you pick from the field's list. When a type with a workflow is assigned, the item becomes a [task](/learn/features/tasks) with state tracking and assignment. ## Task state An item with a task state shows a checkbox next to the title. Click the checkbox or press to toggle between open and closed. See [Tasks](/learn/features/tasks) for full task management. ## Send to Every item has a **Send to** menu in its toolbar, and on meeting tiles, that routes the item without leaving it: - **Start New Chat:** start a chat about the item with Tana, and send it to one of your [custom agents](/learn/features/agents) or a teammate from the Agents and Team sections of the same menu. This creates a new chat with the item attached and opens it, ready for your first message. Choosing a teammate makes a shared private chat with just the two of you; if they do not already have access to the item, Tana offers to grant it. - **Skills and integrations:** run any [skill](/learn/features/skills) attached to the item's type, or a connected integration like GitHub or Linear. If a skill creates an external issue, it links to it once run. - **Pin to meeting:** attach the item to a meeting so everyone in the call can see it. - **Copy Link:** copy a link to the item. Docs also show a **Copy Public Link** row that turns on link sharing (if it is not on already) and copies the public URL in one step. - **Download Markdown:** download the item as a `.md` file, with the title as an H1 heading and the full content in Markdown. In the Team section, the people you have recently shared docs, chats, and canvases with come first, shown with their profile photos. ## Open in chat Right-click a doc and choose "Open in chat" to iterate on it with AI in a split view: proposed edits on the left, an iteration chat on the right. Describe the change you want and the AI updates the proposal in place. Accepting lands the edits on the original doc; rejecting or closing leaves it unchanged. These iteration chats are owned by the doc itself, so they stay out of your main chats list. ## Side panels The toolbar provides access to side panels: - **Table of contents:** Headings from the doc, for quick navigation within long documents - **Backlinks:** All items that reference this doc, grouped by field type or mention context. Backlinks update automatically as other items link to it. - **Changes:** Chronological history of edits showing what changed, who made the change, and when. Click an entry to see details. - **In common:** Items shared between the doc's participants. Only visible when the doc has multiple participants. Backlinks and Changes are pinned open by default. ### Backlinks Backlinks are a key part of how information connects in Tana. When any item links to or mentions this doc, that reference appears in the backlinks panel. Backlinks are grouped by how the reference was made: through a specific field (shown as "Type > Field"), or as an inline mention. ## Location and access The doc toolbar shows two adjacent buttons for location and access: - **Location** ("Lives in") opens a popover for the document's scope (which space or event it belongs to), with an Open button to navigate to the parent. A doc that does not live in a space or on a meeting shows a book icon and a "Library" label, and clicking it opens the [Library](/learn/features/interface#more). - **Access** opens access mode, participants, and Guest Link. See [Spaces and sharing](/learn/features/spaces#access-modes) for the access modes and [Link sharing](/learn/features/spaces#link-sharing) for the Guest Link behavior. The same **Move** action lives in every doc panel's overflow menu (Doc, Canvas, Chat, Skill, Agent, Artifact, Type, Space Home, plus Image, Asset, Audio, Call, and Transcript panels). It is also available on the meeting event overflow menu and on the inline document popover. ## Deleting a doc Delete from the overflow menu or context menu. A confirmation dialog warns before proceeding. A deleted document is soft-deleted: opening it shows a "Document deleted" panel with a "Restore document" button, so you can bring it back if you removed it by mistake. --- #### Home URL: https://tana.inc/learn/features/home Home is the board you land on when you open Tana, with cards for the meeting you are heading into, what came out of recent ones, your tasks, and what you have been working on. - The board you land on when you open Tana: your day, your work, and what came out of your [meetings](/learn/features/meetings) - [Up next](#the-cards) joins a Tana meeting or captures an external one, named for the provider it recognizes - [Recents](#the-cards) brings back what you were last in, split into Jump back in and Recently edited by you - [Work](#work) is the second board: ongoing chats, your tasks, your agents, and your notifications - [Pin any item to a date](#pinning-items-to-dates) and it shows up on the [Calendar](/learn/features/calendar) that day - A [daily brief](#daily-brief) agent can send your day to you in Slack each morning ## The board Home is where you land when you open Tana. It is a board of cards rather than a single list. The set of cards and their layout are fixed. Home shows your own calendar events and organization-level events synced by other members, so a meeting a colleague scheduled on a shared calendar appears alongside your own. ## The cards - **Greeting:** The time-of-day greeting, which flips with the hour, and a short summary of the day. - **Up next:** The meeting you are heading into, with one action on it. For a Tana meeting that is **Join**. For an external one it is **Capture**, named for the provider Tana recognizes, or **Capture anyway** when a colleague is already capturing the same meeting. Where capture cannot run, on the web and on Linux, the button carries the provider's own name and opens the meeting there instead, and a meeting with nothing to enter falls back to **Open**. When meetings clash, every one of them stays reachable rather than the card picking a winner. - **Earlier today:** Meetings that already finished today. - **Meeting outcomes:** What came out of recent meetings, and the follow-ups worth picking up. See [Outcomes](/learn/features/meetings#outcomes). - **Tasks:** Your tasks, with filters. See [Tasks](/learn/features/tasks). - **Pinned today:** Items pinned to today. Drop one on the card to pin it. A row of create buttons at the top of the card starts a doc, canvas, or meeting. - **Recents:** Documents you have been in lately, ranked by how recently and how often you open them. It shows a fixed set in two groups, **Jump back in** and **Recently edited by you**, with **Search** in the card heading. A meeting you opened can appear under Jump back in, which holds what this device opened; Recently edited by you leaves meetings out. Meeting tiles across Home and the other boards show the title, then the AI-written tagline, with the avatar stack below and a strip of up to three teaser cards mixing pinned items, accepted outcomes, and pending proposals, each with a type indicator and a cover thumbnail where it has one. On an external meeting the provider logo carries a frown icon, reading "This meeting is not in Tana. Move it?", on uncaptured events that still have a real provider link. Clicking it runs [Host meeting in Tana](/learn/features/meetings#host-meeting-in-tana). ## Starting a meeting or a capture Two buttons on Home cover both: - **Meet** starts a meeting now, with **Schedule a meeting…** behind it for one at a set time. Both open the same meeting card, where you add people and confirm. See [Creating a meeting](/learn/features/meetings#creating-a-meeting). - **Capture** starts a [system-audio capture](/learn/features/meetings#capture-with-system-audio) for a meeting that is not on your calendar. On the web, or where system-audio capture is not available, it links to the desktop download instead. For anything else, use the create menu (the pencil icon) in the sidebar. See [Interface](/learn/features/interface#create-menu). ## Work **Work** is the second board, for what you owe people rather than what happened: ongoing chats, your tasks, your agents, MCP chats, and your **Notifications**. The sidebar's Work row carries the unread notification count, because the Notifications card lives here. See [Work](/learn/features/interface#work) for what shows up in it. Each space opens on a board of the same kind. See [The space home](/learn/features/spaces#the-space-home). ## Pinning items to dates Any document, task, or other item can be pinned to a date. It then appears on that date, on Home's **Pinned today** card when the date is today, and on the [Calendar](/learn/features/calendar) day. Meetings are not pinned this way, because they appear on their scheduled date automatically. To pin an item, use **Pin to Date** from any document's context menu, by right-clicking the item or opening its toolbar menu. To unpin, right-click a pinned item and choose **Unpin**. Pinning to a date is for work that belongs to a day. For the items you want permanently within reach, use **Pin to sidebar** instead, which files them into your own [pinned area](/learn/features/interface#pinned-items) of the sidebar. Pinned dates are shared, so everyone in the organization sees the same pins on a date. ## Daily brief A daily-brief agent sends you one Slack direct message each morning: the meetings on your day, what happened since yesterday, preparation worth doing, and tasks or follow-ups you may have dropped. It covers the meetings you attend, not everything your account can see. Set it up from the card headed **Get a daily brief in Slack every morning** and press **Set up**. That card lives on the day view at `home.tana.inc/calendar`, so open that address directly to reach it. Dismissing the card hides it on every device. Setup needs your personal Slack account connected, from **Settings**, **Integrations**, **Slack**, or from the card itself. **Link Slack account**, in the **Connected** menu, is an optional second step. It lets Tana ask Slack who you are instead of matching addresses, so it is needed when your Slack email differs from the one you sign in with. The agent is yours once it exists. Open it to change the prompt, its tools, or the schedule, and Tana leaves it alone afterwards. It can message only you. See [Agents](/learn/features/agents#schedules) for changing when it runs. ## No calendar connected With no synced calendar, the **Up next** card carries a prompt to connect one. Connect Google Calendar or Outlook and your events appear automatically; see [Integrations](/learn/features/integrations#calendar-sync) for setup. Dismissing the prompt removes it from Home. --- #### Integrations URL: https://tana.inc/learn/features/integrations Integrations connect Tana to the tools your team already uses: Google Calendar, Outlook mail, GitHub, Slack, Linear, Jira, and HubSpot, reachable by the AI in chat and by skills on your types. - [Google Calendar and Outlook](#calendar-sync): sync events to your Tana calendar - [GitHub](#github): create and manage issues, review PRs, read code from AI chat - [Slack](#slack): send messages and list channels from AI chat, mention @tana in any thread, plus the /tana slash command - [Outlook mail](#outlook-mail): search and read your mail, draft a reply, and send once you accept the action - [Linear](#linear): create and track issues from meetings and chat - [Jira](#jira): create and update issues from meetings and Chat, with meeting screenshots embedded as visual context - [HubSpot](#hubspot): search contacts, companies, and deals from your CRM, log notes, and update lifecycle stages from Chat - [Custom MCP servers](#custom-mcp-servers): connect your own remote MCP servers so their tools are available to AI chat - [Tana Outliner](#tana-outliner): connect Tana to your Tana Outliner graph over its hosted MCP server (beta) - [Coding tools](#coding-tools): send prompts to Cursor, GitHub Copilot, Codex, and Lovable from AI chat - All integrations are accessible to AI in [chat](/learn/features/chat) as tools - Connect from [Settings](/learn/features/interface#settings) with one-click OAuth ## Connecting integrations All integrations are configured in Settings. Click "Connect" to start the OAuth flow and authorize the service. On the desktop app, the authorization page opens in your default browser so passkeys and password managers work; Tana finishes the connection once the provider reports it as active. On the web, authorization opens in a popup. Once connected, the integration is available to AI in [chat](/learn/features/chat) and through [skills](/learn/features/skills). You can disconnect or reconnect integrations at any time from the same settings page. A service that is not listed here can still be reached by Tana's AI as a [custom MCP server](#custom-mcp-servers), if an MCP server exists for it. ### Writing to a connected service Tana's AI reads from a connected service on its own, but it never writes to one without you. Filing an issue, sending a message, or updating a record arrives as an **action** you approve, and the same card then reports how the run went. See [Proposals and actions](/learn/features/proposals#actions). ## Calendar sync Connect **Google Calendar** or **Microsoft Outlook** from Settings, Calendars, to see your events on [Home](/learn/features/home) and the [Calendar](/learn/features/calendar). On the desktop app, connecting or reconnecting opens the provider's sign-in page in your default browser, so passkeys and password managers work. On the web it opens in a popup. Only calendars you own are offered for syncing. One that someone shared with you does not appear here, even if you can edit it, though a shared calendar you already had enabled stays listed so you can turn it off. Each calendar can be enabled, disabled, or force-synced individually. Events sync automatically when changes happen on either side, and you can trigger a manual sync at any time. The first time you connect a provider, Tana enables that account's primary calendar automatically so you have events to look at right away. Reconnect a provider you had connected before and Tana restores the calendars you had enabled, skipping any that no longer exist at the provider. ### Read access first, edits when you grant them Connecting a calendar gives Tana read access, so your day shows up without handing over the ability to change anything. Under **Calendar Accounts**, a read-only account carries **Enable edits**: "Let Tana create and make updates to meetings in your calendar." What **Enable edits** does depends on the account: - **An account connected read-first** gets write consent added to the connection you already have. Your synced calendars are untouched. - **An older account with no read tier recorded** is reconnected instead. Confirm with **Reconnect**, and your synced calendars and settings are kept. Until you grant edits, the per-calendar write settings are inert and note that Tana has read-only access. While a connection is read-only, anything that would write to the provider says so rather than failing quietly. An RSVP, adding a Tana meeting link, moving a meeting, or replacing a series all explain that edit access is missing and offer a way into calendar settings. Ask the AI to create or move a calendar event and it tells you the connection is read-only instead of reporting an unexplained failure. ### Two-way sync Once edits are granted, events you create and manage in Tana sync out: the event appears at the provider shortly after you make it, and later changes to its title or time, and deleting it, propagate too. Choose which calendar receives them by setting a **default write target** in Calendar settings. A per-calendar toggle can add a Tana meeting link to eligible synced events automatically. ### Private events stay out of Tana Events marked **Private** or **Confidential** in Google Calendar or Outlook are dropped at sync time. They never appear in Tana for you or for anyone subscribed to your calendar, and no title, description, or attendee list is stored. To bring a private event into Tana, change its visibility at the provider first. ### When a calendar stops syncing If a connected calendar stops syncing, Tana surfaces a **Calendar needs reconnect** indicator next to your account in the sidebar and on the calendar settings page, and a callout above the day's meetings with a button into calendar settings. The callout is amber with **Open calendar settings** while the provider's token is still valid, and red with **Reconnect calendar** once it has lapsed. The account row also shows a **Sync error** or **Needs reconnect** badge. A successful sync clears all of it. If a connected calendar's authorization expires or is missing a required permission, Tana shows a callout explaining what is needed, with a reconnect action. A calendar Tana still syncs but the provider no longer offers you, one you have lost access to or that was deleted, stays listed with a **Turn off to remove** note, so you can switch it off instead of being stuck with a row you cannot clear. Synced events show their original title, time, attendees, and external meeting link. You can start a chat and extract outcomes from synced events the same way as Tana-created meetings. For more on working with synced events, see [Meetings](/learn/features/meetings#calendar-sync). ## GitHub Connect your GitHub account to work with repositories, issues, and pull requests from AI chat. What the AI can do with GitHub: - **Repositories:** List your repositories and browse directory contents - **Issues:** Search, read, create, update issues. Add comments. - **Pull requests:** List, create, review, merge PRs. Add comments. - **Code:** Read file contents from repositories When a [skill](/learn/features/skills) is attached to a type for GitHub issue creation, you can file an issue from the item's [Send to](/learn/features/docs#send-to) menu; the AI asks which repository when the item does not make it obvious, then links to the GitHub issue once it is created. Filing an issue from a meeting reads the transcript, pulls in screen-share screenshots, and asks a clarifying question if the surface is ambiguous, then attaches an automatic **Screenshots** section to the issue body. ## Slack Slack connects to Tana in two independent ways. They do different things, and neither one enables the other. ### Slack for your organization (admin) An admin installs the Tana Slack app once for the whole organization. This is what lets anyone reach Tana from inside Slack, with no per-person setup: - **`/tana`** starts a 30-minute Tana meeting from any channel or DM. Whatever you type becomes the event's title; with no text, the event is named after the conversation instead (the DM's people, or the channel). Tana posts the event link back in the same place. - **`@tana` in a thread** brings Tana into the conversation. It posts a thinking placeholder and edits it in place with the answer, and keeps a running chat per thread so follow-ups have context. It reads the earlier messages in the thread and any files shared there, and ingests image, PDF, text, and JSON attachments on the mention, so you can drop in a screenshot or document and ask about it. Tana only uses what each participant in the thread is allowed to see. If your Slack email does not match a Tana account, `@tana` and `/tana` reply with an **open in Tana** link that pairs the two identities, so future mentions resolve to you. ### Your own Slack account (personal) Connect Slack from Settings, Integrations to let Tana's AI act in Slack as you. With it connected the AI can: - List channels and users - Send direct messages - Send channel messages Messages the AI sends arrive formatted, and links to Tana documents are rewritten as links your teammates can open. Agents running on a schedule use this same connection, which is how the built-in **Daily brief** skill delivers: ask for a daily brief and it gathers today's meetings and what needs preparing, the people and companies you are meeting for the first time, and your open tasks, follow-ups, and scheduling conflicts, then sends them to you as a direct message. Ask for it explicitly: a general question about your day or your priorities is answered in the chat rather than sent to you as a DM. ## Outlook mail Connect Outlook from Settings, Integrations. Once connected, the AI can: - Search your mail by sender, recipient, subject, unread state, attachments, or when it arrived - Read a message you point it at - Draft a reply for you to send from Outlook yourself - Send a message, once you have accepted the [action](/learn/features/proposals#actions) that proposes it Reading your mail happens in a walled-off step that holds no other tool. It can search and read your mailbox and report back, and it cannot write anywhere, reach your workspace, or send anything. Text inside a message is treated as content to report on, never as instructions to follow. Connecting Outlook mail is separate from [syncing your Outlook calendar](#calendar-sync). Neither one enables the other. ## Linear Connect Linear to create and manage issues from meetings and chat. What the AI can do with Linear: - List teams - Search existing issues - Create new issues - Update issues Like GitHub, Linear has a system [skill](/learn/features/skills) for creating issues from typed items. Attach it to a type and run it from any item's [Send to](/learn/features/docs#send-to) menu. ## Jira Connect Jira to create and update issues from meetings and Chat. Useful for pushing bugs and feature requests out of meeting outcomes and into the project tracker. What the AI can do with Jira: - List Jira projects the connected account has access to - Search issues by free text, optionally scoped to a project key - Create issues with summary, description, type, priority, labels, assignee, and parent (for subtasks or epics) - Update an existing issue's summary, description, priority, labels, assignee, or workflow status Filing a Jira issue from a meeting reads the transcript and embeds screen-share screenshots into the issue description as visual context, the same way it works for GitHub. ## HubSpot Connect HubSpot to read and update CRM data from Chat. Useful for pulling context into a meeting prep, logging a note after a call, or moving a deal forward without leaving Tana. What the AI can do with HubSpot: - **Contacts:** Search by name, email, company, or domain. Look up full details and the deals associated with a contact. Update lifecycle stage, lead status, and any contact property (job title, phone, custom properties). - **Companies:** Search by name or domain. Update lifecycle stage and any company property (industry, employee count, custom properties). - **Deals:** List recent deals sorted by last modified. Create a new deal and associate it with a contact or company. - **Notes:** Create a note and attach it to a contact, company, or deal, for example to log meeting takeaways straight into the CRM record. Property names follow your HubSpot configuration (HubSpot internal property names like `lifecyclestage` or `hs_lead_status`, with the values configured in your account). New to HubSpot in Tana? Follow the [setup guide](/learn/guides/connect-hubspot) for the connect flow, example prompts, and troubleshooting. ## Custom MCP servers Beyond the built-in integrations above, Tana can connect to your own remote [MCP](/learn/features/mcp) servers, so any tools they expose become available to AI chat. This is how you bring a service Tana does not integrate with directly, your own internal server or a third party's, into the AI. To add a server, go to Settings, Integrations, then **Add a tool**. Pick a listed server where one exists, or choose **Custom MCP server**: 1. Enter a name and the server's `https://` URL, then authorize it over OAuth. On the desktop app, the authorization opens in your default browser so passkeys and password managers work. 2. If a server does not let Tana register itself automatically, the connect step says so, and you can add it with a Client ID and Secret instead. Each connected server is private to you; other members of your organization do not see it. Adding your own server is a member permission an admin can switch off, as **Member-added MCP servers** on the organization settings Integrations page. Where it is off you use the servers your admin has approved instead, and servers you added yourself stop being available to your chats until it is switched back on. Connecting a server adds a linked [skill](/learn/features/skills) for it. The AI loads that skill itself when a prompt calls for the server's tools, so you never have to bring it into scope by hand, and a connected server does not carry its tools into every conversation. Manage the connection from the skill's panel: - **Tool catalog:** every tool the server exposes, with the server URL and when it was last synced. Use **Refresh tool catalog** to re-read the list after the server changes. - **Allow or deny each tool:** a per-tool control. A denied tool is never offered to the AI or called on your behalf, and it stays denied across a catalog refresh. Remove a server from the same settings page to disconnect it and drop its tools. This is the inverse of Tana's own [MCP server](/learn/features/mcp), which lets external AI tools connect into Tana. ## Tana Outliner If you also use Tana Outliner, you can connect it to Tana so your Outliner graph is available to Tana's AI chat. Tana Outliner runs a [hosted MCP server](https://outliner.tana.inc/learn/features/local-api-mcp#hosted-mcp-server-beta), also called **Remote MCP**, and Tana connects to it the same way it connects to any [custom MCP server](#custom-mcp-servers). - **Connection type:** HTTP - **URL:** `https://app.tana.inc/mcp` Add it from Settings, Integrations, with **Add a tool** and then **Custom MCP server**: enter the URL and a display name, then authorize over OAuth. Tana Outliner asks you to allow the connection; once you approve, its tools are available in your chats through its linked skill. Once connected, Tana's AI works with your Outliner graph both ways. It can search your nodes, read and navigate the node tree, and read your supertag schemas, so you can pull Outliner context into a Tana document or a meeting, for example asking a meeting's private chat what your Outliner notes say about the topic at hand. It can also write back, creating tagged nodes and todos in a specific place in your Outliner hierarchy, including today's day node, and setting their fields as it goes. The hosted MCP server is in beta and requires a paid Tana Outliner plan. Tana Outliner's local MCP server, which runs inside the desktop app, stays free. New to this? Follow the [setup guide](/learn/guides/connect-tana-outliner) for the connect flow, example prompts, and troubleshooting. ## Coding tools Tana can hand off the current document or chat context to an external coding tool as a self-contained prompt. The hand-off is driven by the built-in **Open in coding tool** skill (see [Skills](/learn/features/skills)) and triggers on phrasings like "create a PR", "open a pull request", or "send this to Cursor". When more than one tool is enabled, the AI asks you to pick. When the repo or project is not named in the source material, the AI asks where the change should land instead of guessing. Available coding tools: - **Claude Code:** Anthropic's CLI coding agent, desktop app only. Configure one or more named projects (name and local folder path) in Settings, Coding Tools. The AI launches Claude Code in the chosen project folder, and the session URL appears as a clickable link in the chat. The web app shows a "Requires desktop app" note. - **Codex:** OpenAI's coding agent - **Cursor:** AI-first code editor - **GitHub Copilot:** GitHub's AI pair programmer (opens in VS Code) - **Lovable:** AI full-stack app builder Enable or disable individual tools in Settings, Coding Tools. Each tool has a toggle to control whether the AI can use it. When a tool is used in chat, an "Open in [tool name]" button appears in the conversation. If the prompt is too long to fit in that tool's launch link, a note appears above the button with a **Copy full prompt** action that puts the complete prompt on your clipboard; the shortened link still works as a secondary action. --- #### Interface URL: https://tana.inc/learn/features/interface Tana's interface is built around a flat sidebar of places you can be, a create menu, quick chat, and search across everything. - Flat sidebar: [Search](#search), [Home](#home), [Calendar](#calendar), [Work](#work), [Spaces](#spaces), and [More](#more) - [Work](#work) holds your notifications, which is why its row carries the unread count - [Pinned items](#pinned-items): pin documents, spaces, agents, and types to your own sidebar, grouped into sections - [Create menu](#create-menu): make meetings, docs, canvases, tasks, chats, types, skills, and custom types from anywhere - [Quick chat](#quick-chat): press Cmd+L to start an AI conversation instantly - [Search](#search) searches text as written, and turns a sentence into filters when you ask it to - Desktop app supports [tabs and split view](#desktop-features) for working with multiple documents ## The sidebar The sidebar is the main navigation in Tana. It sits on the left side of the screen as a flat list of destinations: Search, Home, Calendar, Work, Spaces, and More. Your own pins sit below them, and the account menu at the bottom. There are no tabs. The rows above More are places you can be, rather than a list of things the organization holds. More is the index of everything. On wider screens the sidebar is always visible and can be resized by dragging its right edge. On narrower screens it appears as an overlay when opened. On the calendar and meeting routes it takes a dark theme to match the meeting background. ### Home The [Home](/learn/features/home) board. Where you land when you open Tana: the meeting you are heading into, what came out of recent ones, your tasks, and what you have been working on. ### Calendar The [Calendar](/learn/features/calendar): your meetings as day, week, and month grids. The row stays lit across all three. ### Work **Work** is the board for what you owe people: ongoing chats, your tasks, your [agents](/learn/features/agents), MCP chats, and your **Notifications**. The sidebar row carries a badge with your unread notification count, because the Notifications card lives here. The Notifications card shows its most recent entries, with a **View N more** button for the rest. Notifications cover when someone: - **Assigns you a task.** Assigning a task to yourself does not notify you, including when the AI does it while acting on your behalf. Ask the AI to assign a task to someone else and that person gets a notification attributed to you - **Starts a call** you are waiting on - **Sends you a chat message** - **Archives or restores a type** in your organization. The row carries the type's current title, correct across renames, and links to its panel Each row shows an unread dot before the title, the actor's avatar, and a relative timestamp. Click a row to go to its source, or use **Mark all read** in the card header. **Urgent notifications**, an incoming call being the one that matters, fire outside the app and take you straight to the meeting when clicked: a system notification on the desktop app, and a browser notification on the web. Browser notifications have to be enabled first. The **Notifications** row in **Settings**, **Preferences** shows whether they are on and offers an **Enable notifications** button, or points you to your browser settings if you have blocked them. ### Spaces Your spaces appear in the sidebar so you can jump to any space home in one click. Each space can be expanded to show the types defined in it. See [Spaces](/learn/features/spaces). ### More **More** expands in place rather than opening a page. Under it sit the organization's contents by kind, plus a **Library** row that opens the full index. The Library page shows an ordered preview of each category, **Spaces · Meetings · Tasks · Types · Chats · Agents · People · Skills · Canvases · Docs**, each with a **See all** link to its own page. Empty categories hide themselves. A category page shows the category title, a **Library** breadcrumb to step back out, and an inline search field; section routes (Spaces, Chats, Agents, Skills, Canvases, Docs) also carry the shared **Created by** and **Visibility** filter pills, and Chats adds an **MCP** pill that narrows to chats started from an MCP client. The Types page adds a single-select space filter. The **Skills** section lists only the skills your organization has authored; Tana's built-in system skills are filtered out. The **Meetings** preview defaults to a **Recent** window, with a dropdown to switch it to **Upcoming**, **Today**, or **Past**. It hides itself when no meetings fall in Recent, but once you narrow to an empty window the header stays so you can switch back. Media, meaning images, audio, video, and uploaded files, does not appear in the Library, in space browse, or in Search; it still renders inside the chats and documents that embed it. Anything that does not live in a space or on a meeting lives in the Library, and its location pill reads **Library**. A space home uses the same section index, scoped to that space. ### Tasks **Tasks**, under More, opens the global tasks view: your tasks across every space, grouped by type and narrowed with filter pills (Assigned to, Created by, Type, Space) that live in the URL, so a filtered view is bookmarkable and Back and Forward step through it. It opens on what is assigned to you, and with no filter set it shows your most recent tasks. See [Tasks](/learn/features/tasks) for how the view works. ### People **People**, under More, shows every member and group in your organization as a flat list. A **Filter** narrows it by name, matching a person's full name even when they are displayed elsewhere by a nickname. Each row carries **Meet now**, which creates an instant 10-minute meeting with that person, **Chat**, and **Create**, which opens the create menu scoped to them. Clicking a person opens their **person page**: a time-grouped list of everything the two of you share, with Space and Type filters, plus the same Meet now, Chat, and Create actions in the toolbar. Your own page shows **Edit profile** instead. Drag a task onto a person to assign it to them. ### Pinned items Below More sits your own pinned area: the items you want one click away. It is private to you, and it renders nothing until you pin something. Documents, spaces, [agents](/learn/features/agents), and [types](/learn/features/types) can all be pinned here. To pin, use **Pin to sidebar** from an item's right-click menu or its toolbar menu, or drag the item onto the sidebar. The submenu offers the top level, any section you have already made, and **New section…**, which names a new section and files the item into it in one step. An item sits in one place in this tree, so pinning it to a different destination moves it rather than duplicating it. Top-level pins render as plain rows with no "Pinned" header, and sections group pins under their own header. Drag to reorder pins and sections, to file a pin into a section, or to move one back out to the top level. Right-clicking empty space in the sidebar also offers **New section…**, and sections can be renamed or removed from there. Right-click a pin to unpin it; unpinning drops the pin and leaves the document untouched, so pinning it again restores it. A pin whose document you have lost access to stops rendering. The same collection has a full-page view at `/pinned`, showing top-level pins first and then a group per section, with the same drag-and-drop behavior and a toolbar action to add a section. You can also ask [AI chat](/learn/features/chat) to pin or unpin items for you, to your sidebar, to a [date](/learn/features/home#pinning-items-to-dates), to a [space](/learn/features/spaces) home, or to a [meeting](/learn/features/meetings) everyone in the call can see. Pins apply immediately rather than arriving as [proposals](/learn/features/proposals), because a pin adds no data and is reversible. ### Bottom section The account menu sits at the bottom of the sidebar: the organization avatar and name, with auth and sync indicators beside them. Clicking it opens settings, organization switching, and sign out. A **?** icon next to it opens **Download apps**, **Learn**, **Changelog**, and **Status**, each in your browser. ## Create menu The create menu is accessible from the pencil icon in the sidebar toolbar, or from the "+" button on space home pages and on a date. It is a searchable palette with these options: - **Meeting**, **Doc**, **Canvas**, **Task**, **Chat**, **Agent**, **Type**, **Space**, **Skill** - **Custom types:** Any types you have defined, grouped by space When used from a space context, created items are scoped to that space and the Space option is hidden. Opening it from inside a space and choosing Space creates a subspace nested under it. Tasks open as inline cards rather than navigating away. Everything else navigates to the new document. ## Quick chat Press on the desktop app to open a quick chat window. Type a message, optionally add participants, and press enter. The chat is created and AI responds automatically. Press to open quick chat with voice dictation active. Your speech is transcribed directly into the message input. Quick chat is the fastest way to ask the AI anything without leaving what you are working on. ## Search Open search from the **Search** entry in the sidebar, or with from anywhere, including while editing a document. It opens as a centered dialog, wide enough to show results in the table view, hosting the same browse interface: a search field for full-text search across Tana, the filter bar, and, when you have not typed a query or set a filter, the per-section index. On iPad and other touch devices it opens as a bottom sheet that stays clear of the on-screen keyboard. What you type searches text as written. Matching a person's name tolerates accents and visually similar characters from other alphabets, so "muller" finds "Müller" and "orjan" finds "Ørjan". The same holds for `@` mentions and people pickers. ### Auto-filter An **Auto-filter** button sits at the right of the search field. Press it, or hit enter, and Tana interprets the sentence into the filter pills below, so "meetings with Jane last week" becomes the Meetings kind plus an attendee and a time filter. Press it again to drop the interpretation and go back to the literal search. If the sentence cannot be turned into filters, the button offers **Try again** and your text matches stay on screen. Editing any filter yourself also clears the interpretation and runs the search from the pills alone. ### Related documents Under the direct text matches, a **Related** group lists documents that are about the same thing without matching the words you typed. When nothing matches literally, that group is all you get, and its heading says so. ### Arranging results - **Group** arranges results by **Type**, **Status**, **Assignee**, or one of a type's own member fields. Each group collapses, and a truncated group carries **Show all**. - **Sort** orders them by **Title** (A to Z or Z to A), **Created**, **Updated**, or **Start time**. The table's column headers sort the same way. - **Display** picks the **Displayed attributes**: which of a type's fields each result row shows as a column. ### Filters Search and the Library share one filter bar of round pills. An active pill turns solid and carries an **✕** to clear it. - **Meetings**, **Tasks**, **Docs**, and **Chats:** one-click buttons for the everyday kinds - **Type:** a picker for anything else, including the types you have defined - **Created by:** a people picker, with you pinned to the top as "(you)" and an **Anyone** option to clear it. It takes more than one person - **Visibility:** narrow by who can see an item, **Any**, **Private to you**, **Shared with you**, **Restricted**, **Open**, or **Anyone with the link** for items shared by link Picking a kind reveals its own sub-filters. **Meetings** adds **When** (Recent, Upcoming, Today, or Past) and **Attendees**, a multi-select people picker that narrows to meetings including everyone you pick. **Tasks** adds **Status**, which takes several states at once and reads them back ("Open, Proposed", with **Any status** to clear it), and **Assigned to** (Anyone, You, Unassigned, or a specific member). When the filters narrow to a single type, that type's own fields join the bar as pills you can edit in place, including its [Options fields](/learn/features/types#options-fields), which offer a checklist of the field's values. Without a query, the filters act as a browse mode. Narrow to **Meetings** only and the view becomes a dedicated meetings table (title, the space it lives in, time, attendees, and a lock for private ones), sorted with the most recent first. ### Keyboard navigation Arrow keys move through the dialog as a grid: up and down step through results, left and right cross between the filter pills and the controls on a row. Focus moves in and out of the search field without leaving the keyboard. ### Saved searches A search worth keeping can be saved as its own **Search** document from the filter row, in search, the Library, and a space's home. Reopen it to run it again, always against what is in Tana now, and pin it like any other document to keep it a click away. A saved search stores how you had the results arranged alongside the query, so it comes back grouped and sorted the way you left it. Editing an item from the result list, changing its status or assignee, updates the row in place rather than leaving a stale one behind. The AI can save a search too, and can point you at one that already exists rather than composing a new query. See [Search results in chat](/learn/features/chat#search-results-in-chat). ## Panel title chip As you scroll down any panel, a doc, canvas, chat, skill, agent, meeting, space home, and the rest, a compact title chip appears in the toolbar showing the item's avatar or thumbnail and its entity-type badge. Click the chip to scroll back to the top. ## Panel back navigation Each panel toolbar's back arrow walks that panel's own history, returning to the panel you opened it from rather than the last URL you visited. The desktop app uses **Back** and **Forward** buttons instead, always visible and disabled when there is nowhere to go. ## Settings Open settings with or from the account menu in the sidebar footer. Settings sections include: - **Profile:** Your name, display name, and profile image. - **Billing:** Your plan, AI credit purchases, and the billing portal. A seat on a trial shows that it switches to the Free plan automatically when the trial ends, with a heads-up before it does. - **Calendars:** Connect Google Calendar or Outlook. See [Calendar sync](/learn/features/integrations#calendar-sync) - **Integrations:** Connect GitHub, Slack, Outlook, Linear, Jira, HubSpot, and your own external [MCP servers](/learn/features/integrations#custom-mcp-servers), so their tools become available in chat - **MCP:** Configure AI assistants to connect to Tana - **Coding tools:** Choose which coding tools are available to AI chat - **Preferences:** The **Call Sounds** toggle that gates the join, leave, raise-hand, and chat tones in [meetings](/learn/features/meetings#chat-in-the-call), the **Notifications** row for browser notifications, the **Detect external meetings** and **Enhanced meeting capture** toggles for [external capture](/learn/features/meetings#external-meetings), and a **Tana links** row with a **Reset** for your saved [open-in-desktop choice](#desktop-features). - **Permissions:** On the desktop app, the microphone and system-audio permissions a [capture](/learn/features/meetings#capture-with-system-audio) needs - **Organization:** Name, default content language, and other org-wide settings. A **Logo** row uploads a picture for the organization, replacing the generated initials everywhere it appears, including the account menu in the sidebar, which is what tells them apart if you belong to more than one. Admins edit this section; everyone else sees it read-only - **Members:** Invite and manage team members (admin only). Each person you invite comes in as a Member or an Admin, and an admin can switch anyone between the two later from the dropdown next to their name. Your own row and the organization owner cannot be changed or removed The account menu also lets you switch between organizations, view pending invitations, leave an organization, and sign out. ## Desktop features The desktop app adds features not available in the browser: - **Tabs:** Open multiple documents in separate tabs within one window. A new tab signs in to the organization of the link that opened it, not whichever organization you used last - **Split view:** Work with two documents side by side. Use on any link to open it in the split pane. - **New window:** Use on a link to open it in a new window. - **Toggle sidebar:** Use to show or hide the sidebar (also under View ▸ Toggle Sidebar). The same shortcut works on iPad with a hardware keyboard. In the browser it stays the page's "Save" shortcut. - **Launch at login:** A checkbox in the app menu on macOS, or the File menu on Windows where it reads **Launch at Startup**, that opens Tana when you log in. On by default. - **Open in Tana:** When you open an org-scoped Tana link in a browser and have the desktop app installed, Tana offers to open it in the app, with **Open always**, **Open once**, or **Continue in browser**. Your choice is remembered; reset it from the **Tana links** row in Settings, Preferences, or from the **Always ask where to open** button on the hand-off screen. Meeting links are excluded so guests can still join calls in the browser. - **System tray:** Your day across every organization you are signed into. See [Apps](/learn/features/apps#system-tray). - **Meeting detection:** The desktop app can detect active Zoom, Teams, and Meet calls for [external meeting capture](/learn/features/meetings#capture-an-external-call). The sidebar and panel toolbars dim while the window is not frontmost, and return to full strength when it regains focus. ## Light and dark appearance Tana follows your system appearance and renders in light or dark to match. The whole app switches together, so Home, meetings, documents, and chats always agree, and changing the setting on your machine applies immediately without a reload. ## Sync Tana syncs in real time across all your devices. The sidebar footer shows sync status. A cloud-off icon appears if authentication has expired and you need to sign in again. If a connected calendar silently stops syncing, an amber **!** indicator with the tooltip "Calendar needs reconnect" appears next to the sync status in the user account row; reconnect from [Calendar settings](/learn/features/integrations#calendar-sync) to clear it. --- #### Keyboard shortcuts URL: https://tana.inc/learn/features/keyboard-shortcuts Every shortcut in Tana, including keyboard shortcuts and text input shortcuts. Organized by feature area, with the correct keys for your operating system. ## Meetings ## Chat ## Editing ## Formatting ## Navigation --- #### MCP URL: https://tana.inc/learn/features/mcp Tana provides a Model Context Protocol (MCP) server so external AI tools can read and write your content in Tana. - [MCP server](#mcp-server) lets external AI tools read and write your content in Tana - One-line setup for [Claude Code](#claude-code), connector setup for [Claude Desktop](#claude-desktop), and [any MCP client](#other-mcp-clients) over OAuth - [Tools](#available-tools) cover search, read, create, update, delete, schema, and skills - [Writes create proposals](#how-writes-work) that you review and approve, just like the built-in AI - The server preloads [your identity and space members](#preloaded-context) so agents can act on your behalf - Going the other way? [Connect your own MCP servers](/learn/features/integrations#custom-mcp-servers) so their tools work in Tana's chat ## MCP server Tana provides a Model Context Protocol (MCP) server that lets external AI tools connect to Tana. From a client like Claude Code or Claude Desktop, an agent can read documents, search content, create and update items, and use the same capabilities available in Tana's built-in [chat](/learn/features/chat). The server lives at `https://home.tana.inc/mcp`, and authentication is handled automatically over OAuth. Settings, MCP has a one-click copy for the URL and ready-made configuration snippets. MCP works in both directions. This page covers Tana **as a server**, external AI tools reaching into Tana. For the reverse, connecting Tana to **your own** MCP servers so their tools become available in [chat](/learn/features/chat), see [Custom MCP servers](/learn/features/integrations#custom-mcp-servers) in [Integrations](/learn/features/integrations). ## Claude Code Add Tana to [Claude Code](/integrations/claude) with a single command: ``` claude mcp add tana --transport http https://home.tana.inc/mcp ``` The command and URL are available for one-click copy in Settings, MCP. ## Claude Desktop [Claude Desktop](/integrations/claude) connects to Tana through Anthropic's **connectors** flow. From the Settings, MCP page in Tana, follow the link to Anthropic's "Use connectors to extend Claude's capabilities" guide and add Tana with the MCP Server URL shown on the page. ## Other MCP clients Any MCP client that supports HTTP transport and OAuth can connect to `https://home.tana.inc/mcp`. Point the client at that URL and it goes through the same authentication flow as Claude Code and Claude Desktop. External clients can also point directly at a calendar event URL and pull the full meeting context in a single `readEvent` call: participants, attendees, transcript references, screen-share screenshots, and the `relatedDocs` index of attached artifacts and chats. The bundled `tana-cli` and the URI validator accept the broader set of doc kinds (event, call, transcript, screen-share, artifact, asset, audio, video, canvas, agent, workflow, calendar-subscription) and the `/o/:org/e/:event` and `/o/:org/c/:chat` URL shapes, so you can pass a URL straight from the browser. ## Available tools The server exposes much of the same toolset the built-in AI uses, so an external agent can do many of the same things Tana's own assistant can. Tools are grouped by purpose: **Read** | Tool | Description | | ---------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `searchItems` | Full-text search across Tana. Wildcard listings (`queries: ["*"]`) sort results newest-first | | `semanticSearchItems` | Meaning-based search that finds conceptually related content | | `readItems` | Fetch the full content, fields, and block structure of specific documents. Distinguishes "kind I do not read" from "URI does not resolve" | | `getItemInfo` | Read an item's metadata and status by URI (access, owner, timestamps, proposal state, external IDs) without loading its body content | | `listEdges` | Find relationships between documents in the knowledge graph | | `readTable` | Read tabular data with cell block IDs for precise edits | | `getTypes` | Inspect type definitions and fields so the agent knows the shape of your data | | `readSkill` | Read the definition of a [skill](/learn/features/skills) | | `contextRetrieval` | Gather the relevant context for a natural-language question, the way built-in chat does | | `readEvent` | Read a calendar event with its participants, attendees, and a `relatedDocs` index of every artifact, event chat, and screen-share image attached to the meeting | | `readFullTranscript` | Read the transcript of a meeting recording | | `readScreenShareScreenshots` | List and read the screenshots Tana captured during a meeting's screen shares | | `readScreenshotImages` | Fetch the actual images for specific screenshots from a meeting's screen shares, so the agent can look at them rather than read a description | | `listEvents` | List calendar events | | `listCalendars` | List the user's connected calendars | **Write** | Tool | Description | | ------------------------ | ----------------------------------------------------------------------------------------- | | `createItems` | Create new documents | | `updateItems` | Edit existing documents | | `deleteItems` | Remove documents | | `moveDocuments` | Reparent documents in the graph | | `rename` | Rename a document, with the change going through the proposal review like any other write | | `manipulateTable` | Insert and update rows, columns, and cells in tables | | `createArtifact` | Produce structured artifacts like summaries and notes | | `updateArtifact` | Edit an existing artifact | | `createEvent` | Create a new calendar event | | `updateEvent` | Edit a calendar event | | `generateImageDocuments` | Create image documents from prompts | | `manageTypes` | Create and change types and their fields | | `manageSpace` | Create or modify [spaces](/learn/features/spaces) | | `manageAccess` | Change access control for an item, the agent equivalent of the share menu | | `createSkill` | Author a new [skill](/learn/features/skills) | | `createAgent` | Create a new agent document, with its skills and capabilities | **Proposal review** | Tool | Description | | ------------------ | ------------------------------------------- | | `listProposals` | List pending proposals from a write session | | `approveProposals` | Apply pending proposals | | `rejectProposals` | Discard pending proposals | The exact list and schemas are always available from the server itself via the standard MCP `tools/list` call. ## How writes work Write tools do not change your graph directly. They create [proposals](/learn/features/proposals) attached to a session, and you review them the same way you review changes from built-in chat. This keeps you in control when an external agent is doing real work on your graph. The typical flow is: 1. The agent calls a write tool. The server returns a `sessionUri` and the proposed changes go into that session as pending proposals. 2. You open the session in Tana and review, edit, or discard the proposals. The chat opens with a short recap from the agent of what it did and why, so you have the context to continue from rather than a bare set of tiles. 3. The agent can call `listProposals`, `approveProposals`, or `rejectProposals` if you have told it to, but it should not auto-approve on your behalf. If you have already authorized a write out of band, for example through an explicit instruction to the agent or a CLI flag, the agent can pass `autoApprove: true` on the write call. The server then applies the proposals in the same round-trip instead of leaving them pending. A few write tools do not accept it, because their proposal is the review step itself. Their schema does not list the argument, and the server ignores it for them. ## Preloaded context When a client connects, the server sends instructions that include your identity and the members of your space. This means an agent can resolve "me", "I", and "my" to the right user URI and assign tasks or reference people by name without an extra lookup. ## Custom MCP servers MCP also runs the other way: Tana can connect to your own remote MCP servers so their tools are available to its [AI chat](/learn/features/chat), much like the built-in integrations. You set these up in Settings under Custom MCP servers, authorizing each server over OAuth and choosing which of its tools the AI may use. See [Custom MCP servers](/learn/features/integrations#custom-mcp-servers) in Integrations for the full setup. For example, if you use Tana Outliner, you can connect Tana to its [hosted MCP server](https://outliner.tana.inc/learn/features/local-api-mcp#hosted-mcp-server-beta) (its **Remote MCP**) so your Outliner graph is available to Tana's AI chat, in both directions. See [Connect Tana to Tana Outliner](/learn/guides/connect-tana-outliner) for the setup, or [Tana Outliner](/learn/features/integrations#tana-outliner) in Integrations for the summary. ## Spotting MCP chats A chat session that an MCP client starts shows an "MCP:" prefix on its auto-generated title in your chat list, so it is easy to spot among your other chats. --- #### Meetings URL: https://tana.inc/learn/features/meetings Tana meetings combine video calls, real-time transcription, and AI that turns the conversation into docs, tasks, and decisions, in one place. - Built-in video calling, real-time [transcription](#transcription) in 33 languages, and [screen capture](#screen-capture) - [Capture](#during-the-call) the conversation during the call as typed items, artifacts, and skills, or make a blank doc, canvas, or task - [Present](#presenting) a document, canvas, or your screen on the shared stage, with guests following along - [Capture external calls](#capture-an-external-call) from Zoom, Teams, or Meet without moving the call - [Record a meeting as video](#recording-a-meeting-as-video) alongside the transcript, with everyone told it is running - AI auto-processes meetings when the call ends: summary, action items, and [artifacts](/learn/features/artifacts) - All AI-generated changes go through [proposals](/learn/features/proposals) you review before they are applied - [Guest access](#inviting-people) via link, no Tana account required ## Setting up a meeting ### Calendar sync Connect Google Calendar or Outlook from Settings to see your events on [Home](/learn/features/home) and the [Calendar](/learn/features/calendar). See [Integrations](/learn/features/integrations) for setup details. Synced events appear with their original title, time, attendees, and meeting link. You can start a chat and extract outcomes from synced events the same way as Tana-created ones. Connecting gives Tana read access. Grant edits with **Enable edits** in Calendar settings and sync runs both ways: events you create and manage in Tana appear in your connected calendar shortly after, and edits to their title or time, and deleting them, propagate back to Google or Outlook. Choose which calendar receives them with the default write target in Settings, Calendars. Until you grant edits, anything that would write to the provider, an RSVP included, tells you edit access is missing rather than failing quietly. See [Read access first](/learn/features/integrations#read-access-first-edits-when-you-grant-them). ### Creating a meeting Every meeting you make in Tana comes with a Tana video call link. Start one from wherever you already are: - **Ask the AI.** Tell [AI chat](/learn/features/chat) what you want, "set up 30 minutes with Jane tomorrow afternoon about the launch". A title and a start time are all it needs; the end time, location, description, and who to invite are optional. The meeting arrives as a [proposal](/learn/features/proposals) you accept before it lands, so you see the times before they are real. - **The meeting card**, from **Meet** on [Home](/learn/features/home#starting-a-meeting-or-a-capture) or by dragging a span on the [Calendar](/learn/features/calendar#drafting-a-meeting). One card covers both starting now and scheduling for later. Add people from the picker, your organization's members first, plus any email address including people outside it, then either leave the time on **Now** and click **Start Meeting** to go straight into the call, or set a start and end and click **Create** to put it on the calendar. Starting now also copies the guest link for you. The picker's confirm reads **Invite** when Tana will send real calendar invites and **Add** when it will only add people inside Tana, with a footer explaining which it is. **Myself with screen share** opens a solo session with the camera off, for recording a walkthrough. - **Meet now from People.** Right-click a teammate in the People view, or in one of your constellations, and choose **Meet now** to open a call with them without scheduling anything. - **The create menu.** Select **Meeting** from the create menu (the pencil icon in the sidebar), or the "+" on a date. A meeting made this way starts at the current time rounded up to the next 5 minutes and runs 30 minutes; made from a date other than today, it starts at 9:00 AM. Change the date, time, and duration from the event view. If the call is happening somewhere else, in Zoom, Teams, Meet, or a room with no video link at all, you do not create it in Tana. Capture it instead, and Tana makes the event for you. See [External meetings](#external-meetings). ### Inviting people Open the event's details to add people. Add teammates by name, or type any email address to invite someone outside your organization. As you type, a second section below your organization's members suggests people you have had calendar events with before, so you can add a regular external contact without typing their address again. Because a meeting's attendees sync out to your connected calendar, everyone you add gets a real calendar invite in their inbox, carrying a short meeting link like `meet.tana.inc/gtr-msnp-kwd`. The meeting card shows that same short link, so what you see matches what your attendees got. Longer links to the meeting keep working. Added participants can see the event, join the call, and see the meeting's outcomes. The details panel shows attendees as an avatar stack with controls to add or remove people; only the organizer can change the invite list, and your own avatar sorts first. If you are an admin and an attendee uses your organization's email domain but is not yet a member, an **Invite to organization** control appears at the end of their row, opening Settings, Members with their email prefilled. It does not appear for personal domains like gmail or outlook. A recurring series carries a short link the same way, and it works for guests as well as members, so an external attendee can join any occurrence from the invite. You can also share the guest link from the pre-call screen for quick access without an invite. Guests join with a display name and do not need a Tana account. If a guest opens the link before anyone from your team has joined, they wait in a lobby and are let into the meeting automatically as soon as a Tana user starts the call. Once in the meeting, guests see a resizable three-column layout that mirrors the host: meeting on the left, the live **Meeting Digest** in the middle (read-only, no pause control), and a surface column on the right that shows public documents pinned to the event and approved meeting outcomes. Cards carry the same coloured type badge that the host sees (Bug, Task, Idea, and any other type), plus thumbnails for artifacts. Clicking a card opens the doc in the public viewer. The surface column stays hidden until there is something to show. If you are signed in to a different organization and click a meeting link for one you are not a member of, you land straight in the guest join flow. On the desktop app it opens in a dedicated guest window, leaving your own workspace untouched. Nothing reaches the host organization until you click **Join** on the pre-join screen, which shows that you are joining as **your name · your org** and notes that the meeting is hosted by another Tana organization, so the host sees your real name, photo, and home org rather than an anonymous guest. ### Preparing the meeting Attach the material the meeting is about to the event itself, ahead of the call. In the meeting's **Pinned** section, use **Find and Pin** to attach a document that already exists in Tana, or **Create new** for a blank doc, canvas, or task to write an agenda into. Everyone invited sees what you pin, and pinned documents update live, so a co-organizer can add to the agenda before you start. Pinning ahead of time also shapes what comes out of the meeting. Anything on stage during the call is drawn from the same pinned list (see [Presenting](#presenting)), and when a meeting has a pinned document such as a Product Track or an open issue, [extraction](#after-the-call) treats it as the subject of the meeting and updates it rather than creating a parallel doc. ## The call ### Joining When you open a meeting with a Tana call link, a pre-call device check appears with a camera preview, audio and video toggles, and device settings (camera, microphone, speaker, background blur). The preview is marked with a "Preview" badge so it is clear you have not joined yet. If you do not have a working camera, the preview area shows "No camera detected" instead. Click "Join Meeting" to enter. You can also click **Join** directly on a meeting card on [Home](/learn/features/home) or the [Calendar](/learn/features/calendar). If the call already has five or more people when you join, your microphone starts muted so you do not interrupt; unmute when you are ready. If you have already been in the call once, a **Re-Join** button appears in the meeting toolbar that takes you straight back in, skipping the device check. When someone is waiting for you in a call, Tana notifies you outside the app and takes you to the meeting when you click it: a system notification on the desktop app, and a browser notification on the web. Browser notifications must be enabled first from the Notifications row in Settings under Preferences. ### Controls The meeting toolbar at the bottom of the screen has the standard call controls and the actions you use during a meeting: - **Mute** (): toggle microphone - **Video** (): toggle camera - **Raise hand** (): automatically lowers when you start talking - **Reactions** (): open the reactions popover for the six defaults, the full picker, and search. See [Reactions popover](#reactions-popover) below for keyboard shortcuts. - **Settings:** camera, microphone, speaker, background blur, noise cancellation - **Capture:** turn the conversation into typed items, artifacts, and skills. See [During the call](#during-the-call). - **Present:** put a document, canvas, or your screen on the shared stage for everyone to follow. See [Presenting](#presenting). - **Private Meeting Agent:** opens your private AI side chat for this meeting. See [Chat in the call](#chat-in-the-call). - **More:** an overflow menu with Copy link, **Record meeting** (see [Recording a meeting as video](#recording-a-meeting-as-video)), and Delete. Only the organizer can delete, and the confirmation says that deleting takes the agenda, recording, and chat with it - **Leave** (): end the call for yourself The call shows only the people who are in it. Anyone on the invite list who has not joined is on the meeting card's attendee roster instead, so the stage stays given over to whoever is actually there. A participant with no name on file reads "Unnamed". The same toolbar design is used across the pre-call device check, the guest lobby, and the live call. A muted microphone or camera shows a gray slash rather than a red highlight. On phones, the toolbar is laid out for smaller screens so the controls stay within reach. The meeting uses a two-column layout with a presenter stage: whoever is speaking or sharing takes the main area, and other participants appear alongside. Camera feeds (or avatars) show names and indicators for speaking and hand raise. Side panels during the call give you access to outcomes and shared context with other participants. If your connection drops, Tana automatically attempts to reconnect. You can switch between meetings without dropping the call: jumping to another meeting keeps you connected through the transition. ### Reactions popover The Reactions button in the toolbar opens a popover with the six default reactions, a "+" to open the full emoji picker, and a search. Once the popover is open: - sends one of the six default quick reactions - opens the full picker - Classic emoticons send the matching reaction without leaving the keyboard: `:)`, `:(`, `:D`, `:P`, `:O`, `:/`, `:|`, `:*` The popover stays open between presses so you can stream reactions in a row. Lowercase variants (`:d`, `:p`, `:o`) fall through to search instead of firing a reaction. ### Participant tiles A large call fills the gallery with participant tiles and folds everyone past it into a single "+N more" tile; a smaller window shows fewer tiles, because they stop shrinking past a minimum size. ### Minimized overlay Navigate away from a meeting and the call collapses into a small overlay at the edge of the screen. It shows up to four participant tiles: one fills the overlay, two sit side by side, three or four form a 2×2 grid, and any others collapse into a "+N" count. The tiles prioritize whoever is speaking, anyone with a raised hand, and the active presenter, with screen-shares and camera-off avatars shown first. Click any tile to return to the full meeting view. A chevron toggle flips the overlay between the left and right edges, and the side you choose is remembered. ### Voice agent Tana's **voice agent** is a hands-free conversation partner that joins a meeting as a voice participant. It is available in any internal call: solo meetings as well as multi-participant ones. Start it by saying **"Hey Tana"** or **"Hi Tana"** (with localised equivalents in Norwegian, Swedish, French, Spanish, German, Italian, Portuguese, and Dutch). Say **"Bye Tana"** or **"Goodbye Tana"** to stop it. While the agent is reasoning between your input and its first audio reply, a subtle progress sound plays so the gap does not feel dead; the sound is tied to the **Call Sounds** toggle in Settings, Preferences. Start it from the voice control at the head of the meeting digest, available in any internal call whatever its size. At rest it is a voice icon in a pill; it becomes the agent's avatar once it is engaged. Click it again to end the conversation. In a multi-participant meeting you can also just say the wake word, and Tana briefly mutes the others while the agent listens to you. Once active, the agent takes a participant tile of its own with a **Stop** button. Voice agent is not available on external-meeting captures. When a Tana user activates the voice agent in a meeting with guests, the guests see the agent's tile and hear its audio. Guests get the same auto-mute toast and manual-unmute override that other participants get. Guests cannot activate the agent themselves. The agent follows the meeting, so you rarely have to explain context before asking. It has the day's agenda and the current meeting's details, what is being said, the meeting's pinned items, what is on the shared screen, and the outcomes captured so far. "What have we captured?" is a complete question. Ask it for work and it runs the job in the background while the conversation carries on: looking something up, drafting or updating a document, finding related notes, or acting in one of your connected tools, including your own MCP servers. Several requests can be in flight at once. Each is acknowledged as you make it, and the result arrives in the agent's next reply at a natural pause rather than over whoever is speaking. It can also present. Say "present the risks doc" to put a document pinned to the meeting on the shared stage, and "stop presenting" to take back off a document you put up yourself. It runs the same flow the **Present** button does, so access prompts and guests behave identically, and it knows what is on stage whoever put it there, so you can ask about what everyone is looking at. Documents that are not pinned to the meeting are out of scope, and the agent says so rather than guessing. In a meeting with several people, the voice agent only reads content that everyone present can access, so a private document one person can see is never read aloud to the room. Your voice agent is remembered if you stop it, leave, or reload the page, so it does not reset mid-meeting. ### Chat in the call A meeting has two separate chat surfaces: - **Private Meeting Agent:** A private chat for you and the AI, opened from the **Private Meeting Agent** button in the meeting toolbar. Only you see it, and the AI responds to every message. Use it to ask questions about the discussion, pull up context from Tana, or have the AI create documents without interrupting the room. The button shows when the AI is replying and surfaces an unread indicator if you have closed the popover, so you can leave it minimized and come back to it. The chat persists across reloads and stays with the meeting. - **Chat With Everyone:** A shared chat visible to all attendees, opened from the chat button in the call toolbar. New messages surface as toasts and an unread badge for anyone who does not have it open; clicking a toast opens the chat anchored to that button. A short ding plays for incoming messages when the chat is closed; you can turn this off with the **Call Sounds** toggle in Settings, Preferences. The AI stays quiet by default. Mention @Tana when you want a response, so the human conversation is not interrupted. The chat is attached to the event and remains available after the call ends. When a guest is in the meeting, AI is turned off in this shared chat: @Tana, agent and skill mentions, the `/` command, and the Skill button are all removed, and a callout explains the pause. This keeps your organization's data out of a chat a guest can read. AI stays off for the rest of the meeting even after the guest leaves. Guests can still send and read plain-text messages; mentions, links, attachments, and dictation are member-only and not available in the guest composer. Your private Private Meeting Agent is unaffected. Both use the same composer as [AI chat](/learn/features/chat), so you can reference documents with @, invoke [skills](/learn/features/skills) with /, attach files, and dictate by voice. ### Screen sharing Any participant can share their screen, including guests. The shared screen takes the main view while participant videos move to a smaller strip. Multiple participants can share at the same time. You start a share from the **Present** button in the toolbar; guests have a dedicated screen-share button. Screen sharing is available on desktop and web. On devices that cannot share a screen, such as iOS, the button is hidden. Guests' shared screens are captured into the meeting's screen captures alongside everyone else's, so they appear in the digest and the post-meeting summary. If one of a guest's screenshots fails to upload, the guest sees a notice. ### Presenting Put a document or canvas at the center of the meeting for everyone to follow, without screen sharing. Open **Present** in the toolbar to bring up a palette of everything you can put on stage: - **Screen or Window:** share your screen (see [Screen sharing](#screen-sharing)). - **New:** create a blank doc or canvas, pinned to the meeting and presented right away, for live notes or a whiteboard. - **Pinned**, **Outcomes**, and **Recents:** the meeting's pinned docs, docs created in this meeting, and your recently edited docs, narrowed to the types that can be presented (text and canvas). Type to search across everything else. Picking a document pins it to the meeting and starts presenting. You edit the title and body live while everyone else follows read-only, seeing your caret in a document, or your cursor, viewport, and selections on a canvas. Followers auto-scroll to stay with you (canvases camera-follow your viewport); scroll or pan away and a **Follow** control appears to snap back. Guests in the meeting can be given access to the presented document in one click, so an external attendee follows the same live document. Anyone who joins while you are presenting sees what is on stage. You can also put an item on stage without opening the palette: right-click a pinned item, an accepted outcome, or a pending proposal on the meeting surface and choose **Present document**, or pick **Present** at the top of that item's **Send to** menu. Asking the voice agent works too (see [Voice agent](#voice-agent)). To stop, open **Present** again and choose **Stop Presenting**, which also ends a screen share. The **Present** button stays highlighted while you present. ### Transcription Tana transcribes conversations in real time. Each participant's audio is transcribed separately, so the transcript includes speaker identification and timestamps. Transcription covers 33 languages. It follows your organization's default content language, set in Settings, Organization, Language, and still detects and transcribes the other languages spoken in the call, so a mixed-language meeting transcribes correctly without anyone switching to English. What the AI writes afterwards, including the summary, outcomes, and document titles, comes out in the organization's language whatever was spoken in the room; see [Default content language](/learn/features/chat#default-content-language). The Tana interface itself is English. The transcription indicator in the call toolbar, or the digest, opens the transcript as a peek in place over the meeting, so you can read back without leaving the call; an **Open full transcript** button promotes it to the full panel, which a transcript link or the sidebar takes you to directly. Lines are grouped by speaker under elapsed-time dividers that stay put as you scroll, and text still being recognised appears as it arrives. From either surface you can: - **Select** a line, shift-click for a range, or extend with the arrow keys and Shift - **Copy** what you selected with ; with nothing selected it copies the whole transcript, grouped by speaker. selects every line - **Edit** the transcript: select a segment to delete it, or delete everything from a point onwards Wiping the entire transcript is a panel action, from its **⋯** menu — the peek deliberately leaves managing a transcript to the panel. ### Recording a meeting as video **Record meeting**, in the meeting toolbar's **More** menu, captures video alongside the transcript. The same item reads **Stop recording** while it runs, and it is unavailable while the meeting is [off the record](#off-the-record). While a recording runs, everyone in the meeting can see that the meeting is being recorded, and the digest carries a **Recording** pill. Its popover shows the elapsed time and offers **Stop recording**, so only starting has to come from the More menu. The finished recording plays back on the meeting afterwards. Deleting a meeting deletes its recording with it. Recording state is read off the meeting itself, so asking the AI whether a meeting was recorded gives you the real answer and a link to playback rather than a guess. ### Screen capture When someone shares their screen, Tana automatically captures screenshots at regular intervals. Duplicate frames are filtered out so only visually distinct content is saved, and each unique screenshot gets an AI-generated description. Captures appear in the digest timeline next to the discussion they belong to, not in the transcript, and they are available to AI when extracting outcomes, so the AI can reference what was shown on screen alongside what was said. A screen share's captures sit together as a grid in the timeline. Click one to open it in place, step through the rest with the arrow keys, and use the back arrow for **All captures**. Each offers **Download image** and **Copy image to clipboard** alongside **Delete**, and **Delete all** clears the set behind its own confirmation. A guest viewing the gallery gets a read-only view. ### Deleting a capture To remove a capture you would rather not keep, such as an overlaid chat window or a frame that adds nothing, open it from its timeline row and choose **Delete** in the image viewer. It works during the share and after the meeting, and takes the capture out of the digest, the timeline, and everything the AI reads about the meeting. A guest viewing the gallery cannot delete. ### Off the record Tana transcribes and captures screen content by default during calls. To pause everything, toggle off-the-record mode from the transcription indicator in the call toolbar. This stops transcription, screen-share capture, and external-meeting capture at once, so nothing about the conversation is recorded, and [recording as video](#recording-a-meeting-as-video) cannot be started while it is on. It covers everyone in the call, guests included. An "Off the Record" indicator appears for all participants so everyone knows recording has stopped. Toggle it again to resume. Use this for sensitive discussions, personal topics, or any part of the conversation you do not want captured. ## Outcomes Outcomes are the docs, tasks, typed items, and artifacts that come out of a meeting. Tana lets you capture outcomes while the conversation is happening, and automatically processes the full meeting when the call ends. ### During the call The meeting toolbar's middle controls are **Capture** and **Present**; see [Controls](#controls). **Capture** turns the conversation into outcomes. It opens with a single lookback-window picker, how far back to read: the last minute, last 2 minutes (the default), last 5 minutes, last 10 minutes, last 30 minutes, or the entire meeting. After the call, the picker defaults to the entire meeting instead. From that span, pick what to make: - **Auto** or a specific **type:** "Auto" captures whatever the discussion points to, or pick a type chip (Task, Bug, Decision, and any custom type) and AI creates a structured, typed document from that part of the conversation. When Tana already knows the concept being captured, the in-progress tile uses its title (for example "Capturing iOS scroll bug...") instead of a generic "Capturing..." label. If you later elaborate on a captured item, the proposal updates in place without showing a new card. - **Artifacts:** generate a [rich artifact](/learn/features/artifacts) like a storyboard or slides from the conversation. - **Skills:** run the built-in **Screen share recap** or **Code prompt** skill, or any of your own [skills](/learn/features/skills). A pinned skill becomes a single tile that reshapes with the skill's lifecycle: while it runs the tile shows the skill name and a live progress indicator, and once it has produced proposals the same tile expands into a group stack of those proposals. Click either state to open the full conversation, and use the More or right-click menu for Open chat and Unpin from event. Captured results are shared with the meeting participants by default; the popover footer shows who the result will be visible to before you capture. You can also capture a specific stretch of the conversation: hover a row in the live **Meeting Digest** and use **Capture as…** to open the same picker scoped to that segment's transcript span, or its **⋯** menu to **Copy transcript segment** (the section's transcript, grouped by speaker). To add material yourself, use the **Pinned** section on the meeting surface: create a blank **Doc**, **Canvas**, or **Task**, or search Tana to pin an existing document. Creating a doc or canvas from **Present** also pins it to the meeting and puts it on stage. See [Presenting](#presenting). A **Summary** tile also appears in the right column when the call has wrapped up, so you can generate the post-meeting summary with one click. When anyone starts a capture, a tile shows up for all participants (for example, "Alice is capturing Action Item") so the room knows what's being pulled out of the conversation. If a capture fails, the error surfaces so you know it did not succeed. Pinned imagery, including screen captures and artifact scenes, opens in a full-screen viewer with copy, download, and arrow-key navigation. On the meeting surface, a tile shows a state control inline and an assignee control in the corner for any item that carries a task or workflow state, plain tasks as well as typed items like a Bug, so you can move it to In Progress or hand it to a teammate without opening it. A plain document shows its kind (Canvas, Doc, Meeting), and a typed item keeps its type badge. The meeting also has its own chat surfaces. See [Chat in the call](#chat-in-the-call) for how AI responds in the private Private Meeting Agent and the shared Chat With Everyone. When Private Meeting Agent creates new docs in the meeting, each one gets a **Pin to meeting** action so you can attach it to the event for everyone else to see. ### After the call Once everyone has left the call, Tana automatically generates a tagline and summary for the event, selects representative screenshots from screen shares, extracts action items and decisions as typed items, and creates structured meeting notes. This runs on its own; you do not need to click anything to start it. To run it earlier, a **Wrap up** button appears in the meeting's final minutes (or once it is overdue) and for participants right after leaving, triggering the summary and outcome extraction immediately. When the conversation clearly points a task at a specific teammate, extraction assigns it to that person, and to several people when it was handed to more than one. If who owns it is ambiguous, the task is left unassigned rather than guessed. If you pinned a document to the meeting, such as a Product Track or an open issue, extraction treats it as the main subject of the meeting and prefers updating that document over creating a parallel one. Every call also produces one canonical **summary document**: an AI-written tagline and short summary, a participants byline, a wide illustration whose scene is drawn from the conversation, and sections for Key Takeaways, Topics Discussed, and Next Steps. After the call, the meeting's middle column becomes a **Summary** panel with a **Summary** and **Timeline** toggle, switching between the summary and the topic-by-topic transcript, plus **Share** and **Edit** buttons. The **Share** button opens the same routing menu as **Send to** on any document: copy a link, start a chat, send to a member or agent, run a skill or integration, and **Download Markdown** to save the summary as a `.md` file. If there was not enough conversation to summarize, the panel says "Nothing to summarize" instead of generating one. If anything went wrong while it was being written, Tana retries so the summary always lands. You can re-trigger any of these manually from the outcomes section at any time. Re-running extraction updates existing outcomes rather than creating duplicates. ### Review and use On the meeting surface, captures are grouped into **Outcomes** and a **For Review** list: AI-generated items wait under For Review as [proposals](/learn/features/proposals) until you accept them, and accepted items move to Outcomes. Right after the call the For Review section shows its progress: while Tana is still extracting, an "Extracting items…" row appears above anything already there; when nothing is waiting it reads "You're all done!", and if the meeting produced nothing to review it reads "Nothing to review from this meeting". Once accepted, outcomes are regular documents in the same space as the meeting. You can edit them, assign types, add fields, and work with them like any other document. Each outcome can be shared with specific people or spaces, making it easy to distribute results to people who were not in the call. ## External meetings You do not need to move your calls to Tana to get transcription and outcomes. There are two ways to work with meetings happening in other tools. ### Capture an external call Capture leaves the call where it is and records it from the side, which is what you want when the other participants are not on Tana or you cannot change the meeting link. It runs on the Mac and Windows desktop apps, not Linux, and nothing records until you start it: every capture is one you began. ### Capture with system audio On the desktop app, Tana captures external meetings through your computer's own audio. Because it captures the audio your machine plays, it works across any meeting platform rather than a fixed set, and it is designed to be lighter on your machine and more reliable. It runs on the Mac and Windows desktop apps, and it is not available on Linux. A small always-on-top card opens a couple of minutes before a scheduled meeting so you can act in one click, with pause and stop controls right in it. Its main button depends on the meeting: - **Open & Capture** for an external meeting that has a join link. One click starts recording and opens the link, so a join that gets blocked never leaves you unrecorded. - **Capture Meeting** for an external meeting with no link. - **Join Meeting** for a Tana meeting with two or more people, which takes you straight into the call. Click the card's title to open the meeting in Tana, and **End** stops the capture and takes you to it. You can also start a capture from **Capture** on [Home](/learn/features/home#starting-a-meeting-or-a-capture) for a meeting that is not on your calendar. If someone else in your organization is already capturing the same meeting, the card names them and asks you to confirm before starting a second recording, so it is not recorded twice. **The companion window.** Once a capture is running, that card becomes a fuller window you keep beside your call. It shows what Tana is capturing, carries the meeting's own title, and gives you the meeting assistant right there, so you can create actions and finish work without leaving the call. Drag it by its grip to move it, and its edge to resize it. Its top strip carries **Collapse**, which shrinks it to a one-row strip and back, and **New window**, which opens a Tana window for the organization the recording belongs to without disturbing the recording. It works with no Tana window open at all. System-audio capture needs the operating system's microphone and system-audio permissions. If Tana does not have them, it will not start recording; instead it brings Tana to the front and opens Settings, Permissions, where you can grant microphone and system-audio access and try again. If the meeting is not already on your calendar, Tana creates a matching event for it. In the transcript your own audio is attributed to you. The other participants are named when [enhanced capture](#enhanced-capture) is on, and split out as positional speakers ("Speaker 1", "Speaker 2", and so on) when it is not. **Auto wrap-up:** External captures get the same periodic mid-meeting wrap-ups and section refreshes that in-app calls get, on top of the post-recording wrap-up. The event picks up a tagline, a summary, and outcome proposals without you needing to open it. ### Enhanced capture **Enhanced meeting capture** in Settings, Preferences is on by default on the Mac and Windows desktop apps. It adds two things to a system-audio capture that the audio alone cannot give you: - **Real participant names.** Transcript segments carry the names of the people speaking, instead of positional labels like "Speaker 1". - **Shared screens.** Whatever anyone shares in the meeting is captured into the meeting's screenshots automatically, so you do not have to take them yourself. To do this Tana reads which meeting window is on screen and who is currently speaking in it, which is why on the Mac it needs the operating system's accessibility permission on top of microphone and screen recording. Without that permission the rest of the capture still works, with recording, transcription and wrap-up unaffected, but speakers fall back to positional labels and shared screens are not captured. The expanded capture panel's top strip always shows an eye icon telling you what enhanced capture is doing right now: - A **blue eye** means it is active right now: names and shared screens are being captured. - A **grey eye** means it is on, but no meeting window has been picked up yet. - A **crossed-out grey eye** means nothing is being captured: either you turned enhanced capture off, or you are off the record. Hover it to see which. - An **amber crossed-out eye** means a system permission is still missing. Click the icon for a menu where you can turn enhanced capture on or off, jump to Settings, Permissions, or read about the feature. Turning it off takes effect immediately, including on a meeting that is already recording: no further shared screens are captured and no new speakers are named, while names already matched in that meeting stay. While you are [off the record](#off-the-record) nothing reaches the meeting, neither shared screens nor speaker names, and a blue eye turns crossed-out to say so. ### Detecting a meeting you did not schedule **Detect external meetings** in Settings, Preferences is on by default on the Mac and Windows desktop apps. With it on, Tana notices meeting-like activity from your microphone and audio even when there is no calendar event, and offers to capture it. Tana only prompts for an app it recognizes: Slack, Chrome, Microsoft Teams, Zoom, Discord, Edge, or a generic browser or call. Other audio activity does not trigger it, and a meeting in a browser tab only prompts when your camera is on too. A **Meeting in progress** card appears with **Capture Meeting** and **Dismiss**, headed by the matching calendar event's title when one lines up and **Capture this meeting?** otherwise, naming the app it detected. Dismiss it, or stop a capture while the meeting is still going, and Tana will not prompt you about that meeting again. ### Host meeting in Tana For meetings synced from an external calendar, the event organizer can replace the external video conference link with a Tana call. Click "Host meeting in Tana" on the event. All attendees are notified of the change. **Capture** keeps the call in the external tool and records from the side. Use this when you cannot change the meeting link. **Host meeting in Tana** replaces the link so everyone joins in Tana. Use this when you want the full experience including live outcome capture during the call. --- #### Proposals and actions URL: https://tana.inc/learn/features/proposals AI in Tana never changes your content, or writes into a connected service, without your approval. Every change arrives as a proposal you review, and every write to another tool as an action you accept. - AI never changes anything without approval. Every change is a proposal you review first - [Three types](#proposal-types): create, update, and delete - [Review individually](#reviewing-proposals) with a diff view, or [accept all](#accept-all) at once - [Iterate with AI](#iterating-with-ai) to refine a proposal before accepting - Meeting [outcomes](/learn/features/meetings#outcomes) use the same proposal system - [Actions](#actions) write to GitHub, Slack, Linear, Jira, HubSpot, or your mail, only once you accept them - The action card then [reports the run](#run-status) and offers a [run chat](#the-run-chat) when it needs you - [Version history](#version-history) tracks all changes to documents over time ## Proposal types Every proposal is one of three operations, matching what AI can do to your content: - **Create** (labeled "To be created"): A new document the AI wants to add to Tana. Accept with **Keep**. - **Update** (labeled "To be updated"): Changes to an existing document, shown as a diff against the original. Accept with **Update**. - **Delete** (labeled "To be deleted"): A document the AI wants to remove, shown with a preview of what will be deleted. Accept with **Delete**. Proposals come from [chat](/learn/features/chat), [meetings](/learn/features/meetings), [skills](/learn/features/skills), and any other place AI acts in Tana. Access changes proposed by the AI (restricting a doc to specific participants, opening it back up, or toggling the Guest Link) follow the same flow. The proposal preview shows an access callout with visibility (Participants only, Open, or inherited) and per-participant role changes so the consequences are clear before you accept. ## Reviewing proposals ### Pending proposals When AI generates changes, proposals appear as inline cards on the message that produced them, so you can see what the AI did and decide on it in place. While AI is still working, a progress line shows its status, starting with "Thinking…" and then reflecting what it is doing, for example "Creating 3 items". Cards fill in as they are produced, and the bulk accept and discard actions appear once it finishes. Proposals are grouped on the card by operation (deletes first, then updates, then creates) and by document type within each group. Each pending tile, and the detail toolbar, carries a compact location and access pill, so you can see where the document will land and who will see it before accepting. A move proposal names the destination space. Setting a type on a proposal offers the types relevant to where it is going, including those defined in subspaces and in the Library. Where a proposed typed document would land in a space that conflicts with its type's home space, the card explains the redirect ("This item will be created in 'Product' because the 'Feedback' type is scoped to that space") and accepting lands it in the type's home space rather than failing. ### Accepting and discarding For each proposal: - **Accept** applies the change to Tana - **Discard** rejects the proposal without applying it Click any proposal to open the detail view with the full content of the proposed change. Proposals are editable while pending, so you can modify content directly before accepting. That includes fields while you are comparing versions: the field rows are the same live editors as the document's own, so you can change a value, add a field, or remove one mid-review, with the previous value struck through beside the new one. A proposed task shows its state control and assignee row too, so you can see who it is for without accepting it first. The detail view has previous and next chevrons to step through everything pending, and discarding one advances you to the next, so you can work through a stack without closing and reopening. The view closes once nothing is left to review. If a change can't be applied in full, a colour-coded callout appears above the accept and discard controls explaining what happened: a change that was blocked, one that applied with some parts dropped (for example an access change that could not include a Guest Link for that kind of item), one that has gone stale and should be discarded, or one that hit a temporary problem and will retry. If a proposal was made by a newer version of Tana than the one you are running, the card shows a callout asking you to reload, with a **Reload** button, in place of the Accept and Discard controls. [Accept all](#accept-all) shows the same callout where it reaches one of these proposals. For updates, two controls decide what you are looking at: - **The version picker** in the header selects which document you view, including the original. On a chain that started from a create still waiting to be accepted, that first entry is labelled **Draft**. - **Compare with Original** (reading **Compare with Draft** on a draft chain) overlays the changes onto the version you picked. It is on by default, greyed out for artifacts since there is nothing to diff, and does not appear at all for a deletion or a space change. Asking the AI to edit a document it has only just proposed compares against that first draft, so a stack of edits on something not yet accepted still shows what changed. ### Accept all Click **Accept All** to apply all pending proposals at once. If the batch includes space deletions, a confirmation dialog warns that deleting a space removes all its documents permanently. After acceptance, a confirmation message appears in the chat showing how many changes were accepted and listing the affected documents. ### Iterating with AI When reviewing a proposal from a meeting or outcome, you can chat with AI to refine it before accepting. The detail view opens with a split layout: the proposal on one side, an iteration chat on the other. Describe what you want changed and the AI updates the proposal in place. In meetings, refinements update the same proposal silently, so no new "Capturing..." card flashes back to the top of the meeting. Each iteration creates a new version. Use the version dropdown to compare versions ("Version 1", "Version 2 (Latest)") and accept the one you prefer. ## Outcomes In the context of [meetings](/learn/features/meetings) and background AI processing, proposals are called **outcomes**. The outcomes panel shows: - **Pending outcomes:** Proposals waiting for your review - **Accepted outcomes:** Documents created or updated, grouped by type - **Progress:** Active AI processing with a status indicator showing what it is doing (searching, reading, creating) Each accepted outcome has a **Share** button for managing access independently from the meeting or chat. See [Sharing outcomes](/learn/features/spaces#sharing-outcomes). ## Actions Tana's AI never writes into a connected service on its own. Ask it to file an issue, send a message, or update a record and it drafts an **action** for you to approve. The same card carries the whole thing from proposal to result, so nothing moves or disappears when you accept. An action card shows: - The service it will write to, above the title - The details as `Label: value` chips, the assignee, project, cycle, or whichever fields the target service takes, over the rest of the body - The approve controls at the foot while it is pending. **Accept all** takes a batch at once Sending mail is always this shape. Asking the AI to send an email produces an action you accept, and the message goes out only after you do. Sending is never something a chat can do by itself. ### Run status Once you accept, that same row reports the run rather than handing you off somewhere else: | Status | What it means | | ------------------- | -------------------------------------------- | | **Running…** | The run is in progress | | **Sent** | Everything landed | | **Partly done** | Some of it landed | | **Waiting on you** | The run needs an answer before it can go on | | **Failed** | Nothing landed | | **Needs attention** | The run finished in a state worth looking at | Where the run created something you can open, the card links straight to it. Inside a run, each individual operation reports **Done**, **Failed**, **Waiting on you**, or **Not supported yet**. Accepting an action leaves you where you were rather than navigating you away. ### The run chat A run that is waiting on you, or that did not fully land, offers its **run chat**. That is where you answer what it is blocked on, and where you retry a failure. ### Actions from a meeting An action drafted in a [meeting](/learn/features/meetings) carries the meeting's context with it. The relevant transcript and any screenshots ride along into the record that gets created, so the issue or message arrives with the evidence attached. An action you run from your own private meeting chat stays off the meeting's shared surface. ## Automatic approval Some AI actions skip manual review. During meetings, periodic wrap-ups (tagline, summary, screenshot selection) are auto-approved so the event stays up to date in real time. Document creates from background processing may also auto-approve. Updates, deletes, and space changes always require your approval, regardless of context. ## Version history The changes panel on any document shows a chronological history of edits. Each entry shows what changed, who made the change, and when. Click an entry to see details. Changes are grouped by time and can be expanded to see individual edits. Change types are labeled as Created, Updated, or Deleted with color-coded indicators. Version history is read-only. It does not currently support reverting to a previous version. --- #### Skills URL: https://tana.inc/learn/features/skills Skills are reusable AI instructions you create once and use everywhere. - Reusable AI instructions you write once and use everywhere: [chat](/learn/features/chat), [meetings](/learn/features/meetings), and [typed items](#skills-on-types) - [System skills](#system-skills) are built-in for common tasks like creating GitHub issues - Invoke with [/skills in chat](#invoking-skills), from the [Send to menu](#skills-on-types) on typed items, or from [meeting outcomes](/learn/features/meetings#outcomes) - Write instructions in a rich text editor with full formatting support ## What skills do A skill is a set of AI instructions packaged for reuse. Instead of typing the same prompt every time you want the AI to do something specific, you write the instructions once as a skill and invoke it whenever you need it. Skills are how you customize what the AI does for your team. A product team might create skills for "Write a PRD from meeting notes" or "Create a GitHub issue from this bug." A consulting team might create "Generate a client summary" or "Draft a follow-up email." ## Creating a skill The usual way to make a skill is to ask the AI: describe what it should do, or tell Tana to save something it just did as a skill, and it writes the skill for you. You can also create one yourself by selecting "Skill" from the [create menu](/learn/features/interface#create-menu), and you can open and edit any skill the same way no matter how it was made. However it is created, a skill has three parts: - **Title:** The name that appears in search results and the Send to menu - **Description:** When and why to use this skill. Helps the AI decide whether to use it and helps teammates understand its purpose. - **Body:** The full instructions, written in a rich text editor with formatting support. This is the prompt the AI follows when the skill is invoked. Skills live in [spaces](/learn/features/spaces) and are accessible from that space's context. ## Invoking skills Skills can be invoked from several places: - **Chat:** Type `/` in the [chat](/learn/features/chat) input to search and attach a skill. The AI follows the skill's instructions when responding. - **Send to menu:** Skills attached to a [type](/learn/features/types) appear in the Send to menu on every document with that type. Pick one to run the skill on that document. - **In a meeting:** Open the **"+"** menu in the meeting toolbar and pick a skill under "Rich artifacts and skills" to run it with the meeting transcript and context. See [meeting outcomes](/learn/features/meetings#outcomes). - **@mention:** Type `@` in chat and select a skill to attach it. ## Skills on types When you configure a [type](/learn/features/types), you can attach skills that appear in the [Send to](/learn/features/docs#send-to) menu on every document of that type. This is the most powerful pattern for skills: the skill is always right where you need it. For example, attach a "Create GitHub issue" skill to a Bug type. Every bug document can file one from its Send to menu, and it links to the GitHub issue once created. ## System skills System skills are built-in and read-only. They handle common tasks like creating GitHub issues, creating Linear issues, extracting meeting outcomes, summarizing meetings, generating PRDs and product feedback, and handing off context to coding tools. System skills show a "System Skill" badge and cannot be edited. The **Open in coding tool** system skill packages the current document or chat context as a self-contained prompt and hands it off to Claude Code, Cursor, Codex, GitHub Copilot, or Lovable. It is triggered by phrasings like "create a PR", "open a pull request", or "send this to Cursor". When more than one coding tool is enabled, the skill asks you to pick. When the repo or project is not named in the source material, the skill asks where the change should land instead of guessing. See [Integrations, Coding tools](/learn/features/integrations#coding-tools). You can use system skills as-is or create your own skills that build on the same patterns. --- #### Spaces and sharing URL: https://tana.inc/learn/features/spaces Spaces group related content in Tana with their own members and AI personality. Access inherits from organization to space to document. - Containers for [documents](/learn/features/docs), [types](/learn/features/types), and [skills](/learn/features/skills), each with their own [AI instructions](#ai-instructions) - Three [access modes](#access-modes): everyone, only you, or selected people - Access [inherits](#how-access-inherits) from organization to space to document - [Link sharing](#link-sharing) lets anyone view a document without a Tana account - [Real-time collaboration](#real-time-collaboration) with live presence and cursor sharing - [Nested spaces](#nested-spaces) for deeper organization - Each space opens on [its own home board](#the-space-home) of pinned items, types, meetings, and chats - Custom [color and name](#name-description-color) make each space recognizable throughout the interface ## Creating a space Select "Space" from the [create menu](/learn/features/interface#create-menu). New spaces are private to you by default. Add team members through the location and access settings when you are ready to share, or restrict access further on individual documents inside the space. Opening the create menu from inside a space creates a **subspace** nested under that space. See [Nested spaces](#nested-spaces). ## The space home Every space opens on a home board, the same shape as [Home](/learn/features/home) but scoped to that space. It is both the landing page for the space and the place where you create new content in it. The cards sit in two columns. One column holds: - **Space:** the space itself, its description and configuration - **Types:** the types defined in the space, with a row for making one and **All types** for everything scoped to that space. Click a type to open its [type page](/learn/features/types#the-type-page) - **Create:** Doc, Task, Canvas, and **More** for everything else - **Meetings:** what is next in the space, then what has already happened, newest first The other holds: - **Pinned:** a tile grid of the space's pinned items, with **Find and pin** in the heading - **Ongoing chats:** always scoped to this space, so everyone sees the same list - **Recently edited:** the space's documents by when they last changed ### Pinned items A space's pins are shared and set by hand, so everyone in the space sees the same ones. Use **Find and pin** in the card heading, or drag any item from the space onto the card. Unpin by right-clicking and selecting "Unpin." ### Chat **Ongoing chats** lists the conversations in the space. Starting a new one scopes it to the space, so the AI receives the space's context and follows its AI instructions. See [AI chat](/learn/features/chat). A space home has no search box of its own: [Search](/learn/features/interface#search) in the sidebar covers the whole organization, and **All types** and **See all** links scope the Library to the space. ## Configuration Open "Configure Space" from the toolbar overflow menu on the space home. ### Name, description, color The name and description explain what the space is for. The color is a hue slider that colors the space avatar throughout the interface. Every reference to the space, whether in the sidebar, search results, or document breadcrumbs, shows the same color, making it easy to recognize visually which space content belongs to. ### AI instructions Custom instructions the AI follows when working in this space. This is how you give the AI context about the team, project, or domain. For example, a product design space might say: "This space contains product design documents. When creating tasks, always include a priority field. When writing specs, follow our format: problem, solution, tradeoffs, success criteria." Every chat, skill, and outcome generated in this space inherits these instructions. This is what gives each space its own AI personality. ### Skills Skills created in this space are listed in the configuration for reference. See [Skills](/learn/features/skills) for how to create and use them. ## Nested spaces A nested space is simply a space that lives inside another space rather than at the top level. Create one with the **+** in the top right of the parent space and choose **Space**. Like any item, a space has a **Lives in** location; nesting just sets that location to the parent. The nested space appears in the parent's content list and inherits its access by default, and changing a space's location (from its Location pill) moves it and everything inside it, so a space works as a portable container for its context. Use nesting for a genuine sub-area that has its own focus or members, for example a department space with a sub-space per team. When you just have many of the same kind of thing, reach for a [type](/learn/features/types) instead, so you do not end up with a sprawl of spaces. ## Access modes Every document, space, and meeting has access controls. Open the location and access settings from the toolbar (the lock or participant icon) to manage who can see it. The popover is organized into accordion sections for Scope, Access, People, and Guest link, so each concern has its own place. Three access modes: - **Everyone in {location}:** All members of the organization or space can access the content. - **Only me:** Only you can see it. This is the default for new spaces. - **Only selected people:** You choose exactly who has access. Add people from your organization using the search field, and remove them individually. ## How access inherits Access flows from outer containers to inner content: - **Organization** → all members are added automatically when they join - **Space** → inherits from the organization, can restrict to specific members - **Document or meeting** → inherits from its parent (space or org), can restrict further An item inside a restricted space is visible only to people who have access to the space, even without its own explicit restriction. You can tighten access on individual documents beyond the space's settings, but you cannot open it wider than the parent allows. Every document has a location shown in the location and access popover. You can move documents between spaces from the same popover, which changes both where the document lives and which access rules it inherits. Types can only be moved up to the organization level, not between spaces. ## Link sharing Enable "Anyone with the link can view" in the access settings to create a public link. Anyone with the URL can view the document without signing in to Tana. This is useful for sharing meeting outcomes, docs, or artifacts with people outside your organization who do not need edit access. The public viewer renders typed items with their full type chip, workflow state pill, assignees (as an avatar stack matching the in-app view), fields grid, and embed cards. Embed cards are clickable; embeds for items that do not have their own public page render as plain item cards. The browser tab title reflects the document's title once the page loads. Turning link sharing off applies immediately and the shared URL stops working. If you open a link whose sharing has been turned off, you see a Sign in call to action that returns you to the authenticated view. ## Real-time collaboration When multiple people have access to a document, they can work on it at the same time. Changes appear in real time for all participants. Docs and canvases show live cursor presence with each person's name and color, so you can see who is editing where. [People](/learn/features/interface#people) in the sidebar shows who is online (green dot indicator). Click a person to see everything you share with them. ## Sharing outcomes Meeting [outcomes](/learn/features/meetings#outcomes) and chat results can be shared independently from the conversation they came from. Each outcome has a **Share** button for managing its access settings. This is common when AI extracts action items or documents from a meeting that stakeholders outside the meeting need to see. Share the outcome with them without sharing the full meeting. ## Organization members Invite and manage your team from Settings (admin only): - **Invite new members** by email. An invitation is good for 30 days - **View pending invitations**, each with a copy-link button to re-copy the invite URL, and the option to revoke it - **Manage existing members**, switching anyone between Member and Admin All organization members can be added as participants to spaces, documents, and meetings. [People](/learn/features/interface#people), under More in the sidebar, shows every member. ## Deleting a space Delete a space from the overflow menu on its home page. Deleting a space removes the space and all its contents, including nested spaces. A confirmation dialog warns before proceeding. --- #### Tasks URL: https://tana.inc/learn/features/tasks Tasks in Tana are documents with a state and an assignee, gathered in the tasks view. - Create from anywhere: [create menu](/learn/features/interface#create-menu), [chat](/learn/features/chat), [meetings](/learn/features/meetings), or the [tasks view](#the-tasks-view) - [Tasks view](#the-tasks-view): grouped by type, with bookmarkable Assigned to / Created by / Type / Space filters - Four states: [inbox, in progress, completed, later](#task-states) - [Assign to team members](#assignment) or drag a task onto a person in the sidebar - Toggle state with the checkbox or Cmd+Enter (Ctrl+Enter on Windows) - [Pin tasks to dates](/learn/features/home#pinning-items-to-dates) for time-based planning - Tasks with [types](/learn/features/types) appear on the [kanban board](/learn/features/types#the-kanban-board) ## Creating a task Several ways to create tasks: - **Create menu:** Select "Task" from the [create menu](/learn/features/interface#create-menu). The task opens as an inline card, assigned to you by default. - **Chat:** Ask the AI to create tasks during a [conversation](/learn/features/chat). AI-created tasks appear as [proposals](/learn/features/proposals) for your review. - **Meetings:** Extract tasks from [meeting outcomes](/learn/features/meetings#outcomes) during or after a call. - **Tasks view:** Click the "+" button on the tasks page. - **Typed entries:** When a [type](/learn/features/types) has a workflow, creating an entry of that type creates a task in the inbox state. Tasks created from the create menu and the tasks view open as inline cards (a popover over your current view) rather than navigating away. ## The tasks view The global view is titled **Tasks** in the sidebar and the [Library](/learn/features/interface#more). Instead of tabs, it uses filter pills that live in the URL, so a filtered view is bookmarkable and the browser's Back and Forward walk your filter history: - **Assigned to:** who the task is assigned to. Opening Tasks from the sidebar starts here, set to you. - **Created by:** who authored the task. - **Type:** a specific [type](/learn/features/types), or plain tasks. - **Space:** a specific space. With no pills set, the view shows the latest 50 active tasks rather than an empty screen, with a "Showing the latest 50 active tasks" note; a filter that genuinely matches nothing reads "No tasks". By default tasks are **grouped by type**: plain tasks come first, then each [workflow type](/learn/features/types) under its own heading that links to the type's definition. Every group heading carries a **"+"** that creates a task already matching that group, the right type, state, space, and assignee, folding in whatever filters are active. The header create menu offers a plain Task or any of your workflow types, assigned to you. Each row shows the state checkbox, title, type if any, and the relative updated time; long titles wrap up to three lines rather than clipping. Toggle state by clicking the checkbox, or right-click it for the full state menu (inbox, in progress or custom workflow states, completed, later). Click a row's title to open it in a popover over your current view rather than navigating away. Tasks created from this view default to **unrestricted** access (everyone in the space can see them), and self-assigning a task does not add you to the participants list, so you keep your existing access level instead of being demoted from admin to editor. ## Task states Every task has one of four states: - **Inbox:** New tasks land here. This is the triage state for things that need to be reviewed and prioritized. - **In progress (or custom states):** Active work. When a [type](/learn/features/types) has a workflow with custom states, each custom state represents a stage of progress. - **Completed:** Done. Toggle by clicking the checkbox or pressing . - **Later:** Deferred. For things you want to come back to but not right now. Wherever a task appears, its state shows as a single state icon. The icon mirrors the current state at a glance, and inside a custom workflow it draws a progress pie sized by the state's position in the workflow's ordered states. Click the icon to open the state dropdown, right-click for the context menu, and hover to see the current state's name. On a kanban board, drag cards between columns to change state. In the tasks view, use the checkbox or the state menu on each row. ## Assignment Tasks can be assigned to any team member. Set the assignee from the dropdown below the task title, or **drag a task onto a person** in the [People](/learn/features/interface#people) view in the sidebar. Assigned tasks appear for that person in their tasks view. ## Tasks on Home The **Tasks** card on [Home](/learn/features/home#the-cards) shows your tasks with filters, so what is assigned to you is on the board you land on. Pin a task to a specific date using "Pin to Date" from the context menu. Pinned tasks appear on [Home](/learn/features/home) and on that date in the [Calendar](/learn/features/calendar), alongside your meetings. See [Pinning items to dates](/learn/features/home#pinning-items-to-dates). ## Tasks and types Tasks are documents with state tracking. When you assign a [type](/learn/features/types) with a workflow to a document, it becomes a task automatically. The workflow's custom states become the columns on the kanban board for that type. This means you can have different kinds of tasks (bugs, features, design tasks) each with their own workflow stages, all visible in the same tasks view but manageable on their own kanban boards. --- #### Types URL: https://tana.inc/learn/features/types Types give your documents a defined set of fields, so everything of the same type shares one structure. - Ask [AI chat](/learn/features/chat) to create any type you need, with fields and workflow — no templates required - Define custom types with [fields](#fields) (text, link, date, member, options) to structure any document - [Options fields](#options-fields) carry a fixed list of values and render as chips you can filter on - Add a [workflow](#workflows) to enable status tracking with a [kanban board](#the-kanban-board) - Attach [skills](/learn/features/skills) to a type so they appear as quick actions on every entry - [AI instructions](#ai-instructions) per type guide how the AI works with entries of this type - The [type page](#the-type-page) shows all entries as a list or kanban board - Color-coded [type cards](/learn/features/spaces#types-in-a-space) appear on space home pages ## Creating a type The fastest way to create a type is to ask AI. In any [AI chat](/learn/features/chat), describe what you want to track ("a Bug type with priority, assignee, and steps to reproduce") and the AI will create it, with fields, workflow, and AI instructions. No templates to pick from, no pre-made catalogue. Every type you need, you can just ask for. To create one manually, select "Type" from the [create menu](/learn/features/interface#create-menu). Give it a name, add fields, and optionally enable a workflow. Types can live in a [space](/learn/features/spaces) or at the organization level. ## The type page Every type has a page showing all documents with that type assigned. The page displays differently depending on whether the type has a workflow. **Without a workflow:** A simple list of entries sorted by last updated, with a button to create new entries. An option to "Convert to workflow" is available if you want to add status tracking later. **With a workflow:** A tabbed view with three tabs: - **Active:** A kanban board with columns for each workflow state - **Completed:** Entries marked as done - **Later:** Entries deferred for later Drag entries between tabs to change their state. The "+" button creates a new entry that opens as an inline card. ### The kanban board When a type has a workflow, the Active tab shows a kanban board. Columns represent workflow states. Drag cards between columns to change status. Columns can be resized and reordered. Cards open in an inline popover when clicked, so you can review and edit without leaving the board. Each card also carries an assignee avatar on its trailing edge, with a dashed placeholder when no one is assigned. Click it to open a searchable people picker, type a name, and reassign the card without opening it. The same picker backs the assignee shown in a card's detail view. ### Attribute columns and filters A type's entries table, and its workflow board with the Completed and Later tables, can surface each of the type's fields as its own column and filter. - **Columns:** Member fields render as avatar chips, Options fields as value chips, and Link fields as the referenced titles. Date and text fields flatten to text. Empty cells keep the grid aligned. - **Filter bar:** A pill bar sits next to the search field. **Member** fields open a people picker, **Link** fields a typeahead picker limited to the field's target types, **Date** fields a range picker, and **Options** fields a checklist of the field's values. Filters apply to the entries table and to the workflow view at once, and they live in the URL, so a filtered view is shareable and Back and Forward step through it. - **Display menu:** A **Display** dropdown next to the search field toggles individual columns. Member, link, and date fields are on by default, along with the base columns; free-text fields are off. The AI can filter its own searches on the same fields, so "Bugs reported this month" resolves to one scoped query. See [Search results in chat](/learn/features/chat#search-results-in-chat). ## Configuring a type Navigate to the type configuration page from the overflow menu on the type page. Configuration includes: ### Fields Fields add structured data to every document with this type. Five field types are available: - **Text:** Plain text values - **Link:** References to other documents. You can configure which types are valid link targets and set a custom reverse label (the name shown in backlinks). - **Date:** Date values - **Member:** References to members of your organization - **Options:** A fixed list of values to pick from. See [Options fields](#options-fields) Each field can be configured as single-value or multi-value. Add fields from the configuration page by typing a name and clicking "Add." Remove fields from the dropdown menu on each field. ### Options fields An Options field carries a fixed set of values, so everything of the type is classified the same way. Severity, tier, category, and priority are what it is for. Lifecycle stages belong on a [workflow](#workflows)'s states instead. Define the values on the field in the type editor. **Add value** appends one, and each value can be renamed or reordered. A field with no values yet reads "No values defined on this field." On a document the field offers **Select value**, with a search box once the list is long. Values render as chips in tables, cards, type pages, and published pages, and can be single or multi select like any other field. Renaming or removing a value leaves documents that already carry it untouched. A value that is no longer on the field is grouped under **No longer offered** in pickers and filters, so an existing document keeps reading correctly. You can ask [AI chat](/learn/features/chat) to add an Options field to a type, set a value on a document, or filter a search by one. ### Workflows A workflow adds status tracking to the type. Click "Add Workflow" in the configuration to enable it. Every workflow has four base states: - **Inbox:** Where new entries land by default - **In Progress:** The active state. You can rename it, but it cannot be removed. - **Completed:** Done - **Later:** Deferred Between Inbox and Completed you can add more states, including several custom "in progress" ones. Click "Customise statuses" to add, rename, reorder, or remove the states you add. These become the columns on the kanban board. Remove the workflow entirely with "Remove Workflow" to return the type to a simple list. ### AI instructions Custom instructions the AI follows when working with entries of this type. For example, "When creating bugs, always include steps to reproduce and expected behavior." ### Color A hue slider that colors the type badge throughout the interface. When the type lives in a space, the color is inherited from the space. ### Skills Attach [skills](/learn/features/skills) to a type. Attached skills are available from the [Send to](/learn/features/docs#send-to) menu on every document with this type. This is how you add domain-specific AI actions to your workflow: attach a "Create GitHub issue" skill to a Bug type, and every bug can file one in a click. ## Assigning types to documents Any document can have a type. Assign one from: - The **⋮** menu on a [doc](/learn/features/docs), in the full panel or in a peek ("Set type") - A slash command in the doc title - The [create menu](/learn/features/interface#create-menu) (creating a typed entry directly) A type is not permanent. Open the document in its full panel and its **⋮** menu offers **Change type** to pick a different one and **Remove type** to leave it untyped. That menu is the only place either lives: a peek's **⋮** assigns a type to an untyped document but carries no type option once the document has one, and the title slash command only assigns. Values from the old type are kept rather than discarded. They sit on the document as **leftover fields**, grouped under the type they came from and read-only, and become editable again if you change back. Task state and assignees carry over; a custom workflow state belonging to the old type does not. ## Archiving a type Archive a type from the "..." menu on its type page; a confirmation dialog appears first. An archived type and all its items drop out of every menu, picker, type palette, list, Library view, and search result, including semantic search, and the change takes effect immediately without a reload. Existing items keep their type assignment and show a muted "(archived)" label in place of the normal type badge. The type page shows a callout explaining the state, with an "Unarchive" button that makes the type available again everywhere. To find an archived type again, open **Recently archived** under Organization in Settings. It lists every type you have archived, newest first, with how long ago each was archived. Click a row to open that type's page, where you can Unarchive it. --- ### Help #### Apps and platforms URL: https://tana.inc/help/apps-and-platforms Using Tana on desktop, mobile, and the web. ## Does Tana work on Windows? ## Can I capture external meetings on Windows? ## What desktop features are available? ## Can I use tabs on desktop? ## What is split view? ## How do I see today's meetings without opening Tana? ## Can I open a Tana link in the desktop app? ## Does Tana have a dark mode? ## Is there a mobile app? Not yet. Use Tana in your mobile web browser for now. ## Can I use Tana on the web? ## What are the keyboard shortcuts for navigation? ## What are the keyboard shortcuts for editing? --- #### Canvas and artifacts URL: https://tana.inc/help/canvas-and-artifacts Visual canvases and AI-generated artifacts in Tana. ## How do I create a canvas? ## What drawing tools are available? ## Can I embed documents in a canvas? ## Can I collaborate on a canvas? ## What are artifacts? ## How do I create an artifact? ## What types of artifacts can I create? ## How do I edit an artifact? ## How do I share an artifact? --- #### Collaboration and sharing URL: https://tana.inc/help/collaboration-and-sharing Sharing content, managing access, and collaborating in real time. ## What are the different access modes? ## How does access inherit between items? ## How do I share a link? ## How does real-time collaboration work? ## What are shared outcomes from meetings? ## How do organization members work? --- #### Docs and editing URL: https://tana.inc/help/docs-and-editing Writing, formatting, and managing docs in Tana. ## How do I format text? ## How do I add images and attachments? ## How do I mention another document or person? ## How do I see what links to a doc? ## How do I send a doc to a chat, a skill, or export it? ## How do I iterate on a doc with AI? ## Can I change or remove a document's type? ## How do I delete a doc? --- #### Getting started URL: https://tana.inc/help/getting-started Common questions about getting started with Tana. ## Where do I land when I open Tana? ## What is a space and how do I set one up? ## What types of documents can I create? ## How do I organize Tana with types? ## How do I navigate the interface? ## How do I create new things? ## How do I search for something? ## Where do items that are not in a space live? ## Where do I see my notifications? --- #### Integrations and MCP URL: https://tana.inc/help/integrations-and-mcp Connecting Tana to external tools and using the MCP server. ## How do I connect an integration? ## How do I sync my calendar? ## Does connecting my calendar let Tana change my events? ## Can Tana read and send my Outlook email? ## What can the AI do with GitHub? ## What can the AI do with Slack? ## What can the AI do with Linear? ## What can the AI do with Jira? ## What can the AI do with HubSpot? ## How do I connect my own MCP server? ## What coding tools can Tana connect to? ## Does Tana have an MCP server? ## How do I use Tana with Claude Code? ## How do I use Tana with Claude Desktop? --- #### Meetings URL: https://tana.inc/help/meetings Setting up, running, and getting value from meetings in Tana. ## How do I create a meeting? ## How do I invite people to a meeting? ## How do I prepare an agenda before a meeting? ## How do I join a call? ## What happens to a call when I navigate away? ## Can I share my screen? ## Does Tana transcribe meetings? ## Can I capture my screen during a meeting? ## Can I delete a screen capture I did not want? ## How do I pause recording during a call? ## What can I do during a call? ## Can I chat during a meeting, and does AI work with guests present? ## What can I do with a meeting after the call? ## Can Tana record a meeting as video? ## Can someone from another organization join my meeting? ## Can Tana record Zoom, Teams, or Google Meet meetings? ## Can I talk to Tana by voice during a meeting? --- #### Organizing your spaces and types URL: https://tana.inc/help/organizing-your-spaces Spaces, types, fields, and structuring your information in Tana. ## What is a space home? ## How do I configure a space? ## Can spaces be nested? ## How do I create a type? ## How do I add fields to a type? ## How do I make a field with a fixed set of choices? ## Can I filter a type's entries by its fields? ## What are workflows? ## Can I see a kanban board? ## How do I add AI instructions to a type? ## Can I save a search I keep running? ## How do I archive a type? ## How do I delete a space? --- #### Working with AI URL: https://tana.inc/help/working-with-ai How to use AI chat, proposals, skills, and voice input in Tana. ## How do I start an AI conversation? ## What can AI do in Tana? ## Can the AI search Tana and show me the results? ## How do AI proposals work? ## Can AI apply changes automatically? ## Can the AI file an issue or send a message for me? ## How do I know whether an action worked? ## Can I see previous versions of AI changes? ## How do I choose an AI model? ## Can I use voice to talk to AI? ## What are skills? ## How do I create a skill? ## How do I use a skill? ## Can I customize my own AI agent? ## Can the AI change who has access to a doc? ## What language does the AI write in? ## What happens when I hit my AI usage limit? --- #### Your day and tasks URL: https://tana.inc/help/today-and-tasks Managing your day and tasks in Tana. ## Where do I see my day? ## What is on the Home board? ## How do I see my meetings for the week? ## How do I navigate between dates? ## How do I pin items to dates? ## How do I start a meeting or a capture from Home? ## What is the Work board? ## Can Tana send me a daily brief in Slack? ## How do I create a task? ## What are task states? ## How do I assign tasks to people? ## How do tasks appear on Home? ## How do tasks connect to types? --- ### other #### Bring every meeting into Tana URL: https://tana.inc/blog/bring-every-meeting-into-tana Capture meetings from any tool, keep your calendar in sync both ways, and pin your own sidebar. Plus a sharper meeting view and a calmer voice agent. This update is about Tana fitting the way you already work. Your calendar now syncs both ways, capturing the meetings you hold in other tools is smoother and more reliable, and the sidebar is yours to arrange. The meeting itself, and the voice agent in it, got sharper too. Events you create and manage in Tana now sync out to your connected calendar. Make an event in Tana and it appears in Google Calendar or Outlook moments later, and changes to its title or time, and deleting it, propagate back too. New connections write Tana events to your primary calendar automatically. If you connected a calendar earlier, turn it on by choosing which calendar to write to in [calendar settings](/learn/features/meetings#calendar-sync). Capturing a meeting held somewhere else now works better. Because it records the audio your computer plays, it works across any meeting platform rather than a fixed set, and it is lighter on your machine and more reliable. A small always-on-top window opens a couple of minutes before a scheduled meeting so you can start recording in one click, with pause and stop controls right there. You can also start a capture from the Today page for a meeting that is not on your calendar. If a teammate is already capturing the same meeting, everyone else sees it so it is not recorded twice. It is rolling out as an opt-in. On the latest desktop app, turn on **Use experimental meeting capture** in Settings, Preferences. See [capturing external meetings](/learn/features/meetings#capture-with-system-audio) for the details. The sidebar is now yours to shape. Instead of listing every space and agent automatically, it keeps a clean set of primary links and a **Pinned** area you fill with the things you actually reach for. Pin any document, space, or agent, drag to reorder, and right-click to create a new section and group your pins. Your full set of spaces is still there in the Library whenever you need it. The meeting view is now one calm column that carries the work instead of a stack of documents. What the AI captures during the call is sorted into two clear places: items wait under **For Review** until you accept them, with keep and reject buttons right on the card, and accepted ones move to **Outcomes**. During the call the live digest steps back to a slim status line, so the shared focus takes center stage, and the full timeline stays one click away. When the meeting ends you get an editable, shareable summary with a Summary and Timeline toggle. See [how outcomes work](/learn/features/meetings#review-and-use) for the details. ## More in this update The voice agent has a calmer, more ambient look and clearer audio, and you can set a default voice for Tana's built-in agents in Organization Settings. Invite anyone to a meeting straight from Tana. Add teammates by name or type any email address, including people outside your organization, and everyone you add gets a real calendar invite in their inbox. Transcription follows your organization's language across meetings, external capture, the voice cursor, and dictation, while still picking up other languages spoken in the room. Guests in a Tana meeting can ask the AI questions in their own private chat, separate from the shared meeting chat. It answers but cannot create or change anything. Say "Hey Tana" and give a command in the same breath, like "Hey Tana, file that as a bug", and the agent acts on it instead of just greeting you. Tana opens faster, especially on slower connections and right after an update. Incoming-call notifications now reach you in the browser, not just the desktop app, and take you straight to the meeting. The full list of what shipped lives in the [changelog](/changelog). ## Try this --- #### Bring your own tools URL: https://tana.inc/blog/bring-your-own-tools Tana reaches further this update: its AI can use the tools on your own MCP servers, the Tasks view groups and filters your work into bookmarkable slices, and a new search popover gets you to anything in Tana from wherever you are. This update is about reach. Tana's AI can now use the tools on your own servers, a reworked Tasks view groups and filters your work into slices you can bookmark and share, and a new search popover gets you to anything in Tana from wherever you happen to be. Tana already connects to GitHub, Slack, Linear, Jira, and HubSpot. Now you can connect anything else that speaks the Model Context Protocol, your own internal server or a third party's, and its tools become available to Tana's AI. Add a server from settings and authorize it once over OAuth. Each connection becomes a [skill](/learn/features/skills), so its tools only enter a chat when that skill is in scope, and you can allow or deny each tool from the skill's panel. Your connected servers stay private to you. It runs both ways, too. Tana is itself [an MCP server](/learn/features/mcp), so Claude, Cursor, or any MCP client can work with Tana from the other side. See [Custom MCP servers](/learn/features/integrations#custom-mcp-servers) for the full picture. The [Tasks](/learn/features/tasks) view is now built around filter pills, assignee, creator, type, and space, that live in the URL, so any slice of your tasks, everything assigned to you, what you have delegated, only your open issues, becomes a link you can bookmark or share, and the browser's Back and Forward walk your filter history. Group the list by type, status, or assignee, with a "+" on every group heading that adds a task already matching that group and your active filters. With nothing filtered, the view shows your most recent tasks rather than an empty screen. [Search](/learn/features/interface#search) now opens as a popover right where you are, including while you are writing in a document. Start typing for full-text search across Tana, or open it empty to browse. When nothing is typed, the popover shows the [Library](/learn/features/interface#more) as a browsable index: each kind of thing, meetings, tasks, types, chats, and the rest, in its own section with a "See all" page. It is the fastest way to land on something without losing your place. Reach for it with Cmd+P on macOS, or Ctrl+P on Windows and Linux. ## More in this update **Filter by who can see it.** A new visibility filter in [search and the Library](/learn/features/interface#search) narrows any list to what is private to you, shared with you, restricted, open, or shared by link. **Set an [outcome's](/learn/features/meetings#outcomes) state from the meeting.** Outcome tiles show a state control inline for anything with a workflow, a plain task or a typed item like a Bug, so you can move it along without opening it. **See where any item lives.** The location button on a panel now names the [Library section](/learn/features/interface#more) an item belongs to, like "Meetings" or "Docs", and links straight there. **Recent meetings, front and center.** The [Meetings preview](/learn/features/interface#more) on your Library and space home leads with recent meetings, with a dropdown to switch between recent, upcoming, today, and past. **[Reopen an archived type](/learn/features/types#archiving-a-type).** A "Recently archived" list in settings keeps every type you have archived, so you can find one and unarchive it. **A [title chip](/learn/features/interface#panel-title-chip) when you scroll.** Scroll down any panel and a compact chip with the item's icon and type appears in the toolbar. Click it to jump back to the top. **[Dictation](/learn/features/chat#voice-input) keeps your last words.** Send a dictated message the moment you stop speaking and the tail of what you said now makes it in, instead of getting dropped. **Copy a long prompt for a [coding tool](/learn/features/integrations#coding-tools).** When a hand-off prompt is too long for a tool's launch link, Tana offers a "Copy full prompt" button so you can paste the whole thing. **Open a [video](/learn/features/docs#images-and-attachments) in its own player.** Opening a video item now plays it in a dedicated player panel with the usual toolbar, instead of a generic view. **[Transcription](/learn/features/meetings#transcription) spells your team's names right.** Meetings and dictation now lean on your organization's name and your teammates' names, so "Tana" and the people in the room come through correctly. That is the gist. The full list, every new, improved, and fixed item, lives in the [changelog](/changelog). ## Try this The new search popover and Library are about getting to anything fast. Let the AI do the gathering: drop this into a chat and it pulls a scattered topic into one place. --- #### From conversation to action URL: https://tana.inc/blog/from-conversation-to-action The AI can turn a conversation into a calendar event, guests can join the shared meeting chat, and a new Capture as action grabs the exact part that mattered. Plus Meet now or later from Today, diagrams inline in chat, and more. This update is about the conversation, and everything you want to do with it. The AI can turn it into a calendar event, guests can join the shared chat, and you can grab the exact part of a meeting that mattered without keeping the rest. Ask in [chat](/learn/features/chat) and the AI can now [propose](/learn/features/proposals) a calendar event. Give it a title and a start time and it drafts the rest: an end time, time zone, location, description, and the people to invite. Adding participants keeps the event to that list. You stay in control. It is a proposal you review before anything lands, and the AI still will not change the time, title, attendees, or meeting link of an event that already exists. Meetings have a shared chat, Chat with everyone, and now guests are in it too. They can send and read messages right alongside your team, the simplest way to bring an outside collaborator into the conversation as it happens. To keep your organization's data private, the AI steps out of the shared chat whenever a guest is present, with a short note explaining the pause. Your own private meeting agent keeps working as usual. A meeting can cover a dozen different things. Capture as lets you act on just one of them: hover any row in the live [digest](/learn/features/meetings#transcription) and turn that section into a note, a task, a typed item, or an [artifact](/learn/features/artifacts) like a screen-share recap. Scoped to that one part, the AI works from focused context, so what comes out is about that decision or action, not the whole conversation. Need the exact words instead? The same hover menu copies that segment of the transcript, grouped by speaker. The Meet button in [Today](/learn/features/home) splits in two. Meet now starts a meeting straight away and opens a "Who are you meeting with?" popover, where you can copy a guest link, add colleagues in one click, share your screen solo, or start a voice chat with an agent. Meet later opens the scheduling popover instead, for picking a time. One button, two clear intents: start talking now, or put it on the calendar for later. ## More in this update **Diagrams render in [chat](/learn/features/chat) replies.** When the AI includes a diagram, it now renders inline once it is complete, instead of flashing raw markup as it streams. **Open a [task](/learn/features/tasks#the-tasks-view) in a peek.** Click a task's title in the Tasks view and it opens in a popover over the list, so you can read or edit it without leaving your tasks. **A date chip on an event's location pill.** It reads Today, Tomorrow, Yesterday, or a short date, and clicking it jumps to that day on the calendar. **Know when a [type](/learn/features/types#archiving-a-type) is archived or restored.** A notification tells you when someone in your org archives or restores a type, with a link to its panel. **Filter any [Library](/learn/features/interface#more) section.** Created by and Visibility pills now sit on every section page, saved in the URL so the view is bookmarkable. **Your own [skills](/learn/features/skills), front and center.** The Skills section and search now list the skills you and your team have made, with Tana's built-in ones filtered out. **Open org links in the [desktop app](/learn/features/apps).** Click a link to your org's Tana in a browser and it asks whether to open in Tana or stay in the browser, then remembers your choice. Set it from Preferences in the web app. **Find events you organized in [Created by](/learn/features/interface#filters).** Calendar events are now attributed to their organizer, so filtering by Created by surfaces the meetings you set up. The throughline this time: less time spent capturing the meeting, more spent acting on it. The [changelog](/changelog) has the rest of this release, including a handful of call and sign-in fixes. ## Try this The AI can schedule a meeting and prep your team for it in one go, drawing on what it already knows about your work. Drop this into a chat and it drafts the event and a prep doc for you to approve: --- #### Tana auto-detects external meetings URL: https://tana.inc/blog/tana-auto-detects-external-meetings Tana notices when you are in a Zoom, Teams, Meet, or Slack call that never made it onto your calendar, and asks whether to capture it. One click turns it into a transcript, a summary, and outcomes. Plus presenting a real document instead of your screen, Tana in Slack, and our SOC 2® Type 2 report. The meetings that go uncaptured are rarely the ones on your calendar. They are the call someone pulls you into, the Slack huddle, the Zoom link in a DM. Tana now notices when you are in one and asks whether to capture it. One click turns it into a transcript, a summary, and outcomes you review in Tana. Presenting a document no longer means sharing your screen, and Tana answers questions in the Slack thread where they came up. A call you hold in Zoom, Microsoft Teams, or Google Meet takes one click to capture, and no longer starts with a setting. It records through your computer's own audio on the Mac and Windows desktop apps, so the list does not stop at those three. Any tool your machine plays sound through works the same way, and the call stays where it is. The meeting does not have to be on your calendar at all. Tana notices meeting-like activity from your microphone and speakers and asks, so the call someone pulled you into is one click from being captured rather than gone. It only ever offers. Nothing records until you press **Capture Meeting**. [What it looks for](/learn/features/meetings#detecting-a-meeting-you-did-not-schedule), and how to turn it off, are in the docs. For the ones that are on your calendar, a window still opens a couple of minutes beforehand, and **Open & Capture** starts the recording and opens the meeting together, so a blocked pop-up never leaves you sitting in an unrecorded call. Pause and stop live in that window, and for a call that appears out of nowhere you can start a capture from Today. Pick a document or canvas from the **Present** palette in the meeting toolbar and it takes the center of the meeting for everyone. It is not a picture of a document. It is the document, rendered on each person's own screen. That difference is the whole point. Text stays sharp and selectable at any window size instead of arriving as a compressed video of your monitor. People can scroll ahead to read something again and press **Follow** to come back to you. Nobody has to ask whether they can see your screen, and you are not hunting for the right window while everyone waits. You edit it live while the room watches it change, so the notes you take during the meeting are simply the meeting's notes, in Tana and attached to the event the moment the call ends. Guests read the same live page as everyone else. What was on screen survives the call too. The digest records the document you presented, so reading back through a meeting shows which one was on the table at which point. In a Tana call, screenshots from a shared screen sit next to the discussion they belong to. See [how presenting works](/learn/features/meetings#presenting). Meetings produce work that points somewhere else, and following one of those threads used to mean losing your place. Right-click most things in Tana now, a meeting tile, a pinned document, an accepted outcome, and open it in a new tab. The desktop app adds two more places to put it, the side panel beside what you are already reading or a window of its own. Modifier keys do the same without the menu, and on desktop a number key jumps straight to a tab. Search opens as a centered dialog now, wide enough to read results as a table. Not every question is worth opening another tab for. Mention **@tana** in a Slack thread and it answers in place, reading the thread and the files shared in it, so you can drop in a screenshot and ask about it without moving anything. Type **/tana** in a channel or a DM and you get a Tana meeting link back in the same place, so a thread that has outgrown typing becomes a call. It goes the other way too. Give an agent a morning schedule and its work arrives in Slack looking like something a person wrote. Links, bold, bullets, and headings survive the trip, and a link to a Tana document arrives as something your teammates can click. The two directions are set up separately. An admin installs the Tana app once for the organization, which is what **@tana** and **/tana** run on, and you connect your own Slack account in Settings, Integrations to let your agents post as you. [The Slack docs](/learn/features/integrations#slack) have both. A type's fields now show up as columns on its entries table and its workflow board, and as a row of filter pills beside the search box. Narrow proposals to one customer, tasks to one owner, or anything dated to a range, by picking values rather than writing a query. The filters live in the URL, so a filtered board is a link you can send someone, and opening an entry and coming back leaves the filtering exactly as you left it. The AI reads the same fields. Ask for "open proposals from customer Nordfish" or "feature requests in Product area Billing" and it resolves the constraint in one scoped search instead of listing every candidate and narrowing them one at a time. ## More in this update The transcript lives in one place, grouped by speaker under time markers that stay put as you scroll. Click a line or shift-click a range to copy it, or copy the whole thing with nothing selected. Opening it during a call keeps you in the meeting. Reviewing what the AI proposed and correcting it are the same screen. Edit a value, add a field, or remove one while the old version stays struck through beside the new. The meeting wrap-up leaves a field blank rather than guessing at it, and quotes only what was said in the meeting it is summarizing. A record that looks complete gets trusted, so it has to be. Meeting links are now `meet.tana.inc` plus a short code. Short enough to read out on a call, and invites carry the same one the meeting card shows. Anyone can put an agent on a schedule, including agents shared with them. Schedules stay private to whoever made them, and the opening message is optional. Ask the AI to find one of your own agents by what it does, like "find the agent that does customer research". Attach a skill to a message with a slash and it runs straight away, using whatever you typed as its input. Admins can change a member's role between Member and Admin after inviting them, without starting over. Screen shares get their own featured region during a call, and clicking one enlarges it. Raised hands come to the front of the tiles. A stalled calendar now says so, with a banner above Today and a button that takes you where you can reconnect. Large PDFs attach to a chat without failing, so a long report reads like any smaller file. With GitHub connected, the AI can list your organization's members and their roles, and who has access to a repository, so you can ask who owns something before you assign it. **Send to Linear** and **Send to Jira** create the issue now, instead of reporting themselves unavailable. Device settings name the microphone the call is really using, and plugging in a headset mid-call moves the call onto it. Synced events keep their end times, and all-day events pick up later changes, even when Tana cannot identify the organizer. Copying a link while a document is open in the preview overlay copies that document, not the page behind it. Admins can upload an organization logo under Settings. It replaces the generated initials everywhere the organization appears. AI chat spends fewer tokens on the same work, so you are less likely to meet a usage limit. The full list of what shipped lives in the [changelog](/changelog). ## SOC 2® Type 2 Your security team's favorite question has a one-word answer now, and the word is yes. Prescient Assurance LLC examined the controls behind Tana, and the SOC 2 Type 2 report is ready. Request it on our trust center at [trust.tana.inc](https://trust.tana.inc), along with our security practices and the current sub-processor list. Tana Outliner is not in scope. DPA requests go to privacy@tana.inc. ## Try this Capture picks up the meetings you did not plan for, which means the loose ones now leave a trace too. Ask Tana what came out of them. --- #### Tana takes it from here URL: https://tana.inc/blog/tana-takes-it-from-here Tana does more on its own: meetings wrap themselves into outcomes, there's one place to create without leaving the call, and your tasks and mentions show more at a glance. Plus a unified Library, Mermaid diagrams, and Outlook calendar sync. This update shares a thread: Tana does more on its own. Meetings wrap themselves into outcomes the moment everyone leaves, there's one place to create or capture without leaving the call, and your tasks and mentions show more at a glance. Once everyone leaves a call, Tana runs the [wrap-up](/learn/features/meetings#outcomes) on its own: it proposes outcomes and writes a summary, with no manual step. Want them while you are still in the call? Hit Wrap up. When you pin a document to a meeting, the wrap-up treats it as the strongest signal of what the meeting was about, and updates that document rather than creating a new one. The meeting toolbar is simpler. Capture, Skills, and Create collapse into a single create menu, with a spyglass beside it to find and pin things. From that one menu you can turn the last minute, or the whole meeting, into a typed item, make a private note or task just for you, or run a skill or summary. One place, less hunting. Browse is now Library, the home for everything that is not pinned to a space, and it shares one filter bar with [Search](/learn/features/interface). Filter pills jump you to meetings or chats, or narrow by who created an item, and your spaces and types each get a dedicated section so they are easy to reach. Once you have found something, the new ["Send to"](/learn/features/docs#send-to) menu routes it anywhere: start a chat about it, run a skill, send it to an integration, pin it to a meeting, copy its link, or export it as Markdown. A task now shows a state icon wherever it appears, filling like a pie as it moves through its workflow, so you can tell where something stands without opening it. Give any [type](/learn/features/types#workflows) a workflow to make it task-like. Inline mentions carry more too: a task's state, a person's avatar, an image thumbnail, or the item's type. Hover one for a preview, or right-click to pin it to a date, assign it, or change its status. You can now add diagrams to any document. Type "/mermaid", ask the AI in chat, or set a code block's language to Mermaid, then switch between a live preview and the source. When the AI writes a diagram, it self-corrects unsupported syntax on its own. ## More in this update **Connect your Outlook calendar.** Microsoft and Outlook calendars now sync into Tana, so those meetings show up alongside the rest. **Activity is now a dashboard.** Promoted into the sidebar, it shows your notifications and recently updated items side by side, with an unread count. **A guided welcome session for new team members.** New workspaces open with a live, guided meeting: an agent tuned to your work sets up spaces and types with filled-in examples, and walks you through the essentials. **[Agents](/learn/features/agents) know what time it is.** Agents are now timezone-aware, so "today" matches your own calendar day and a [scheduled routine](/learn/features/agents#schedules) set to 09:00 in Oslo lands on the right Oslo morning. **Reassign right from the board.** Kanban cards carry an assignee avatar. Click it for a searchable people picker and reassign without opening the card. **Paste a link, get a mention.** Pasting a link to a Tana item now drops in an inline [mention](/learn/features/docs#mentions), not a raw URL. **Meeting summaries open with an illustration.** A wide pen-and-watercolour image, drawn from what the meeting was about, now tops each summary. **Muted by default in larger meetings.** Join a call that already has five or more people and you start muted, so you don't cut in. **[Archive a type](/learn/features/types#archiving-a-type).** Archive a type and it drops out of every menu, picker, and search right away, while existing items keep it with a muted "(archived)" label. **[Restore a deleted document](/learn/features/docs#deleting-a-doc).** A soft-deleted item now shows a "Restore document" button, so you can bring back something removed by mistake. That is the gist. The full list, every new, improved, and fixed item, lives in the [changelog](/changelog). ## Try this Diagrams are the easiest new feature to try yourself. Drop this prompt into your chat and see what Tana draws: --- ### Guides #### Build a shared issue tracker with a type URL: https://tana.inc/learn/guides/track-issues-with-a-type Create an Issue type so your team tracks work the same way: shared fields, one board, and the AI filing new issues in, complete and consistent, from a meeting or a chat. {/* DRAFT for review. Teaches the structure side via an Issue type. Collaboration-first; value woven through the teaching. */} Every product and tech team tracks issues: bugs to fix, changes to make, things that need doing, each with a priority, an owner, and a state. In Tana you turn that into one shared tracker by giving issues a **type**. Create an Issue type once, and every issue shares the same fields and shows up on one board, whether a teammate files it by hand or the AI files it from a standup. This guide sets that up: the type, the board, and letting the AI keep it fed. ## Create the Issue type Ask the AI to "create an Issue type," or add one from the create menu, in the [space](/learn/features/spaces) your team works in so everyone shares it. A [type](/learn/features/types) is for a kind of thing you have many of, and an Issue is the textbook case. Add the fields an issue always needs: status, priority, the area it affects, and an assignee. Every Issue from now on carries these, so nothing arrives half-described. Give the type a workflow, a set of states like Triage, In progress, and Done. With a workflow, the type's page becomes a [kanban board](/learn/features/types#the-kanban-board): move an issue from Triage to Done and everyone sees it move. ## Let the AI file issues in, complete The point of the type is that new issues land already in shape. When one comes up in a [meeting](/learn/features/meetings) or a [chat](/learn/features/chat), the AI creates an Issue with the fields filled, status, the area, who raised it, as a [proposal](/learn/features/proposals) you review and accept. Nothing lands until you say so. To make this dependable, set the type's **AI instructions**: "capture what needs to happen and why, set priority from how it is described, and for a bug include steps to reproduce." Now every issue the AI files matches how your team writes them, and you stop chasing people for the missing details. ## See every issue in one place Open the Issue [type page](/learn/features/types#the-type-page) and you have every issue across the team, as a list or as the board. Filter to all the open ones, or just the ones for one area, with nobody assembling a report. Because the type lives in your team's space, it is the same board for everyone: the shared source of truth for what is open and who has it. ## Add a one-click action Attach a [skill](/learn/features/skills) to the Issue type and it becomes a one-click action on every issue: "file this in GitHub," "post it to the team's channel." The skill does the same job the same way every time, so sending an issue upstream is one click for anyone on the team, not a copy-paste ritual. (Filing to GitHub or Linear needs the [integration connected](/learn/features/integrations).) --- #### Capture meetings from Zoom, Teams, and Meet URL: https://tana.inc/learn/guides/capture-external-meetings Capture your Zoom, Teams, and Meet calls with the Tana desktop app, with no moving the call or asking anyone to switch, and get the transcript, screenshots, and outcomes in Tana to share with your team. {/* DRAFT for review. Flow verified against feature/meetings.mdx (External meetings). Pitched above the in-flux capture UX; honest troubleshooting drawn from real failure modes in the User insights space. Screenshot deferred until capture UX settles. */} Most of your calls happen with people who are not in Tana, customers, partners, candidates, on Zoom, Teams, or Meet. You do not have to move those calls or ask anyone to switch tools to get them into Tana. The desktop app captures them from the side, so the transcript, the screenshots, and the outcomes land in Tana just like a call you hosted there, ready to share with your team. It is a one-time setup, then a click per call. ## Before your first capture Capture runs in the Tana **desktop app**, so install and open it if you have not. Then, once: - **Grant permissions.** Tana needs microphone and system-audio access so it can hear the call. If either is missing when you start a capture, Tana opens Settings, Permissions so you can grant it and try again. - **Leave Detect external meetings on** in Settings, Preferences. It is on by default and lets Tana offer to capture a call that is not on your calendar. ## Capture a call Join your Zoom, Teams, or Meet call the way you always do. You do not move it or change the link; Tana works from the side. A couple of minutes before a scheduled meeting, a small always-on-top window opens with the meeting title. Click **Open & Capture** to start recording and open the join link together, or **Capture Meeting** for a meeting with no link. For a call that is not on your calendar, Tana notices the meeting activity and shows a **Meeting in progress** card with **Capture Meeting**; you can also start one from the Today toolbar. Either way Tana creates a matching event for it. Then go back to your call and run it normally. {/* TODO screenshot (deferred until capture UX settles): the capture window with the **Open & Capture** button. */} Tana captures the audio for transcription and takes screenshots of shared screens as you go, and it identifies speakers and lines them up with the transcript. There is no bot in the call and no video recording, just the words and the shared screens. For anything sensitive, stop the capture to pause it. When the call ends, or you click **End capture**, Tana processes the conversation into a summary, the outcomes it found, and screenshots from what was shared, all as [proposals](/learn/features/proposals) you review. The same live wrap-ups you get in a Tana call happen here too, so the event fills in without you opening it. ## Capture, or move the call to Tana Two ways to handle a call that lives in another tool: - **Capture** records from the side and leaves the call where it is. Use it when you cannot change the link, or when the other people are not on Tana. - **Host meeting in Tana** replaces the meeting's video link with a Tana call: the organizer clicks **Host meeting in Tana** and attendees are notified. Use it when you want the full in-call experience, live outcome capture, the meeting agent, and shared surfaces. ## If a capture does not start **No capture window or card.** For a scheduled meeting, check that the calendar it lives on is connected. For an unscheduled call, check that **Detect external meetings** is on in Settings, Preferences. **The capture produced no transcript.** This is almost always permissions. Recheck microphone and system-audio access in Settings, Permissions, then capture again. **The call did not show up right away.** Synced calendar events can take a moment; give it a beat or refresh. **Tana won't start a capture.** A permission is missing. Tana brings itself to the front and opens Settings, Permissions; grant microphone and system-audio access, then try again. --- #### Catch up on your team's work, then act on it URL: https://tana.inc/learn/guides/catch-up-and-act Behind after time away or back-to-back meetings? Ask Tana what your team decided and what is on you, then turn it into filed follow-ups in the same chat, each one reviewed as a proposal before it lands. {/* DRAFT for review. Teaching exemplar: the steps are the lesson. Each one teaches a transferable Tana skill (scoping a retrieval prompt, verifying grounded answers, chat as a command interface, the propose/review/accept loop), not just the procedure for catching up. */} Coming back from a few days away, or surfacing from a week of back-to-back meetings, you hit the same wall: the team kept moving and now you have to reconstruct what happened. Tana can do that reconstruction for you, and then act on it without you leaving the conversation. What you are really learning here is what a Tana [chat](/learn/features/chat) is for. It is two things at once: a way to **retrieve** across everything your team has done, and a way to **act** on what you find. Catching up is just a good first place to see both. The same moves work anywhere in Tana. ## Catch up, then act Open a [chat](/learn/features/chat) and ask in plain language: _"What did the team decide this week, and what is waiting on me?"_ Look at what that one question is doing, because it is the pattern behind every good retrieval prompt. _"What did the team decide"_ sends Tana searching your shared record, the meetings held while you were out, the decisions captured from them, the issues raised. _"waiting on me"_ narrows that to the items that mention or are assigned to you. Scoping by **time** (this week, since Monday) and by **relationship** (on me, for my team, from Acme) is how you turn a vague ask into a focused answer instead of a wall of everything. The more precisely you scope, the sharper the reply. The answer comes back with its receipts. Each point is assembled from real items, a specific meeting, a logged decision, an issue, and every one is a link you can open to see the full context the one-line summary was drawn from. Make opening a few of them a habit. Tana builds the answer from what your team actually recorded, and only from what your permissions let you see, so it is grounded rather than guessed. But a thirty-second skim of the sources is what turns a plausible-looking summary into one you are willing to act on. Learning to trust and verify an AI answer this way is a skill that pays off everywhere in Tana, not just here. Here is the shift that makes this more than a digest. Stay in the same chat and tell Tana what to do about what you found: _"Draft replies to the two decisions that need my input, file the bug from Tuesday's standup as an Issue, and add a task to review the migration by Friday."_ Two things to notice. You can stack several actions into one message, and Tana carries the context from the catch-up straight into each one, it already knows which bug, which decision, which migration. The chat that just searched for you is now creating and updating on your behalf. That is the second half of what a chat is: not only "what happened" but "do something about it", in one continuous thread. Nothing you just asked for is applied on its own. Each change comes back as a [proposal](/learn/features/proposals): a reply drafted, an issue to be created, a task to be added. A proposal shows you exactly what will happen, the full body of the new issue, the precise edit to a doc, before any of it is real. You edit what is off, accept what is right, and discard the rest. This is the most important thing to carry out of this guide: the propose, review, accept loop is the same everywhere AI acts in Tana, in [meetings](/learn/features/meetings), through [skills](/learn/features/skills), here in chat. Learn the control once and it holds for all of it. And because your team shares this record, accepted work is live the moment you keep it: the issue lands on the team's board, the task shows up assigned, with no separate step to circulate it. ## Make it a habit Catching up is a natural Monday-morning move, or the first thing to do after time away. Once the shape of the ask is familiar, you do not have to keep typing it: hand it to an [agent](/learn/features/agents) on a [schedule](/learn/features/agents#schedules), and the catch-up, with its follow-ups already drafted, is waiting for you each morning to review. The same retrieve-then-act pattern runs forward too, to [prepare for what is coming up](/learn/guides/prepare-for-meetings). --- #### Connect GitHub URL: https://tana.inc/learn/guides/connect-github Set up the GitHub integration so the AI can file and manage issues, review and merge pull requests, read code, and hand work to Copilot, from your meetings and chat. Tana connects to your GitHub account so the AI can turn what you discuss into filed issues, work with pull requests, and read your code, without switching tools. Talk through a bug or a change in a meeting or in chat, file the issue while the context is fresh, and hand it to Copilot to start a fix. This guide covers the one-time setup, what to expect from the authorization flow, and how to use the integration in chat. ## What you can do with GitHub in Tana Once GitHub is connected, ask the AI in chat to: - **File issues**: create a structured issue from a discussion, with a title, body, labels, and assignees - **Update issues**: change the title, body, labels, assignees, or the open or closed state - **Search and read issues**: look up issues by keyword, useful for checking duplicates, and read an issue with its comments - **Work with pull requests**: list, create, review (approve, request changes, or comment), and merge (merge, squash, or rebase) - **Comment**: add a comment to an issue or a pull request - **Read code**: browse a repository's directories and read file contents at any branch, tag, or commit - **Include screen-share screenshots**: screenshots shared on a call are embedded in the issue body for visual context - **Hand work to Copilot**: assign an issue to GitHub Copilot so it can start a fix and open a pull request You can also attach skills to Tana types that file GitHub issues one-click from your documents. See [Skills](/learn/features/skills) for how that works. ## Connect GitHub You only need to do this once. The connection is per-account, not org-wide. 1. Open Settings with , or from the account menu in the sidebar. 2. Open **Integrations**. 3. Find **GitHub** under personal integrations and click **Connect**. 4. A GitHub authorization window opens. Sign in if you aren't already, and authorize access. 5. Review the access GitHub lists for Tana, then approve it. 6. The window closes and Tana shows **Connected** next to GitHub. ## Permissions Tana requests Tana asks GitHub for the access it needs to work with your issues, pull requests, and code: reading repositories, issues, and pull requests, and creating or updating issues and pull requests on your behalf. Tana never reads or writes anything outside what you approve during connect. Your access token is stored securely on Tana's servers, never in your browser. You can review or revoke Tana's access at any time from your GitHub account settings, under connected applications. ## Use it in chat Once GitHub is connected, the AI can call GitHub tools when a prompt asks for it. Some prompts to try: - "Check the acme/web repo for open issues about the login bug, then file a new one if there isn't a duplicate." - "Create a GitHub issue in acme/web for the bug we just discussed, label it a bug, and assign it to Copilot." - "Read the README in acme/web and summarize the setup steps." - "What open pull requests in acme/web are waiting on my review?" - "Approve PR #142 in acme/web and merge it with a squash." The AI picks the right GitHub tool from your prompt and asks for clarification when something is ambiguous, for example which repository. ## From issue to pull request GitHub is more than an issue tracker in Tana. A common loop: 1. **File a well-scoped issue** from the meeting. The AI can read the repository first to ground the issue in the real files, and checks for duplicates when you ask. 2. **Hand it to Copilot.** Assign the issue to Copilot when you file it, and GitHub Copilot can start a fix and open a pull request. 3. **Review and merge from chat.** Ask the AI to review the pull request, approve it, request changes, or comment, and merge it with a merge, squash, or rebase when it's ready. The AI never assumes the repository. If you don't name one, it asks, and can list your repositories so you can choose, before filing. ## How Tana files an issue A few things happen when the AI creates an issue: - **Confirms the repo**: it never guesses the target repository. If you haven't named one, it asks and can list your repositories so you can choose. - **Duplicate check**: when you ask it to, the AI searches existing issues first, so you don't file the same thing twice. - **Structured body**: the AI writes the body in GitHub-flavored markdown, with context, current and expected behavior, and steps to reproduce, quoting the reporter where it helps. - **Screenshots as context**: screenshots shared on a call during the meeting are embedded in the issue body. ## Disconnect GitHub To remove the connection from Tana: 1. Open Settings → **Integrations**. 2. Open the **Connected** menu next to GitHub and choose **Disconnect**. The same menu has **Reconnect**, which you can use to refresh access. Disconnecting stops the AI from reading or writing to GitHub but does not change anything already in GitHub or in Tana. To fully revoke Tana's access from GitHub's side, also remove Tana from your GitHub account's connected applications. ## Troubleshooting **The AI says it can't reach GitHub.** The connection has likely expired or been revoked. Reconnect from Settings → Integrations. **The AI keeps asking which repository.** This is by design. It never guesses where to file. Name the repository in your prompt, or pick from the list it offers. **A merge didn't go through.** Branch protection rules, required reviews, and merge conflicts are enforced by GitHub. Resolve them on GitHub, then ask again. **Something else?** See [Integrations](/learn/features/integrations) for the broader integration overview, or reach out at [support@tana.inc](mailto:support@tana.inc). --- #### Connect HubSpot URL: https://tana.inc/learn/guides/connect-hubspot Set up the HubSpot integration so Chat can search contacts, companies, and deals, log notes to CRM records, and update lifecycle stages without leaving Tana. Your customer conversations happen in Tana; your customer record lives in HubSpot. Connect the two and the AI moves between them for you: it pulls the contact, company, and deal history into [meeting prep](/learn/guides/prepare-for-meetings), and pushes what the call produced, the notes, the new deal, the stage change, back to the CRM while the context is fresh. This guide covers the one-time setup, what to expect from the OAuth flow, and how to start using the integration. ## What you can do with HubSpot in Tana Once HubSpot is connected, ask the AI in [chat](/learn/features/chat) to: - **Find contacts and companies**: search by name, email, company, or domain - **Read deal context**: list recent deals or pull a contact's deals into a meeting - **Log notes**: drop meeting takeaways onto a contact, company, or deal as a HubSpot note - **Create deals**: open a new deal and attach it to a contact or company - **Update lifecycle stage and properties**: move a contact from Lead to MQL, set a lead status, or write to any contact or company property You can also attach a [skill](/learn/features/skills#skills-on-types) to a type, so a HubSpot action like "log this to the deal" becomes one click on every item of that kind, the same for everyone on the team. ## Connect HubSpot You only need to do this once. The connection is per-account, not org-wide, so each teammate connects their own, and the AI acts in HubSpot with your access. Open Settings with or from the account menu in the sidebar, then open **Integrations**. Find **HubSpot** in the list and click **Connect**. {/* TODO screenshot: Tana Settings → Integrations panel with the HubSpot "Connect" button highlighted. */} A HubSpot authorization window opens. Sign in if you aren't already, and pick the HubSpot account you want to connect. {/* TODO screenshot: HubSpot account picker, the screen that appears when a user has access to more than one HubSpot account. */} Review the permissions HubSpot lists for Tana, then click **Connect app**. {/* TODO screenshot: HubSpot OAuth consent screen showing the scopes Tana is requesting. */} The popup closes and Tana shows **Connected** next to HubSpot, along with the account name. {/* TODO screenshot: Tana Settings → Integrations panel showing HubSpot in the connected state with the account name visible. */} If you have access to more than one HubSpot account, you'll pick one while authorizing. You can disconnect and reconnect later to switch accounts. ## Permissions Tana requests Tana asks HubSpot for the access it needs to read and write the records the AI works with: contacts, companies, deals, and notes. Tana never reads or writes anything outside the scopes you grant during connect. You can review or revoke Tana's access at any time from your HubSpot account under **Settings → Integrations → Connected Apps**. ## Use it in chat Once HubSpot is connected, the AI calls HubSpot tools whenever a prompt asks for it. Some prompts to try: - "What's the latest on the Acme deal? Pull the contact details and the open deals." - "Move Sarah Chen's lifecycle stage to MQL and set her lead status to Connected." - "Log this meeting as a note on the Acme deal and tag it to Sarah." - "Create a new deal called Acme expansion, attach it to Acme Corp, and set the stage to Discovery." {/* TODO screenshot: a Tana chat showing the AI calling a HubSpot tool, ideally the "log meeting note" prompt with the response card visible. */} The AI picks the right HubSpot tool from your prompt and asks for clarification only when something is ambiguous (for example, which Sarah, or which pipeline). ## Working with custom properties HubSpot fields use internal property names like `lifecyclestage` and `hs_lead_status`, and the values for those fields come from how your HubSpot account is configured ("MQL", "Customer", "Opportunity", and so on). If a prompt uses display names ("Marketing Qualified Lead"), the AI maps them to your account's values when it can, and asks if it can't. To update a custom property reliably, name the property and the value the way they're configured in HubSpot. ## Disconnect HubSpot To remove the connection from Tana: 1. Open Settings → **Integrations**. 2. Click **Disconnect** next to HubSpot. Disconnecting stops the AI from reading or writing to HubSpot but does not delete any data already in HubSpot or in Tana. To fully revoke Tana's access from HubSpot's side, also remove the Tana app under **Settings → Integrations → Connected Apps** in HubSpot. ## Troubleshooting **The AI says it can't reach HubSpot.** The connection has likely expired or been revoked. Reconnect from Settings → Integrations. **The AI created a record but I can't find it.** Records the AI creates are scoped to the HubSpot account you connected. If you have multiple HubSpot accounts, double-check you're looking at the right one. **A property update was rejected.** HubSpot rejects property updates when the value isn't on the configured option list (for example, writing "MQL" to a `lifecyclestage` that only has `lead`, `marketingqualifiedlead`, etc.). Use the value as it's configured in your HubSpot account. --- #### Connect Jira URL: https://tana.inc/learn/guides/connect-jira Set up the Jira integration so the AI can create and update issues from your meetings and conversations, and search projects and issues for context. Tana connects to your Jira account so the AI can turn what you discuss into filed and updated issues, without you switching tools. Talk through a bug or a task in a meeting or in chat, and the ticket gets created while the context is fresh. This guide walks through the one-time setup, what to expect from the authorization flow, and how to start using the integration in chat. ## What you can do with Jira in Tana Once Jira is connected, ask the AI in chat to: - **Create issues**: file a new issue from a discussion, with summary, description, type, priority, labels, assignee, and parent - **Update issues**: change fields on an existing issue, or move it through your workflow to a new status - **Search issues**: look up existing issues by keyword, useful for checking duplicates before you file a new one - **List projects**: see the projects you have access to, so issues land in the right place - **Attach meeting screenshots**: include screenshots shared on a call as links on the issue, so the ticket keeps its visual context You can also attach skills to Tana types that drive Jira actions one-click from your documents. See [Skills](/learn/features/skills) for how that works. ## Connect Jira You only need to do this once. The connection is per-account, not org-wide. 1. Open Settings with , or from the account menu in the sidebar. 2. Open **Integrations**. 3. Find **Jira** under personal integrations and click **Connect**. 4. An Atlassian authorization window opens. Sign in if you aren't already, and authorize the Jira site you want to connect. 5. Review the access Atlassian lists for Tana, then approve it. 6. The window closes and Tana shows **Connected** next to Jira. ## Permissions Tana requests Tana asks Atlassian for the access it needs to work with your issues and projects: reading projects and issues, and creating or updating issues. Tana never reads or writes anything outside what you approve during connect, and it does not request billing, admin, or organization-management access. Your access token is stored securely on Tana's servers, never in your browser. You can review or revoke Tana's access at any time from your Atlassian account settings, under connected apps. ## Use it in chat Once Jira is connected, the AI can call Jira tools when a prompt asks for it. Some prompts to try: - "Check Jira for open issues about the login bug, then file a new one if there isn't a duplicate." - "Create a Jira issue in the ENG project: write up the bug we just discussed, set priority to High, and assign it to me." - "Move ENG-412 to In Review and add a note about what changed." - "What projects do I have access to in Jira?" The AI picks the right Jira tool from your prompt and asks for clarification when something is ambiguous, for example which project, or which of two similar issues. ## How Tana files an issue A few things happen automatically when the AI creates or updates an issue: - **Duplicate check**: when you ask it to, the AI searches existing issues first, so you don't file the same bug twice. - **Formatted descriptions**: the AI writes the description in Jira's native format, so headings, lists, code, and links render properly in the issue. - **Screenshots as context**: if the issue comes from a meeting, the AI can attach screenshots shared on the call as links in the description. To change status, the AI uses your project's workflow transitions. It can only move an issue to a status your workflow allows from where the issue is now. ## Disconnect Jira To remove the connection from Tana: 1. Open Settings → **Integrations**. 2. Open the **Connected** menu next to Jira and choose **Disconnect**. The same menu has **Reconnect**, which you can use to refresh access or switch to a different site. Disconnecting stops the AI from reading or writing to Jira but does not change anything already in Jira or in Tana. To fully revoke Tana's access from Atlassian's side, also remove Tana from your Atlassian account's connected apps. ## Troubleshooting **The AI says it can't reach Jira.** The connection has likely expired or been revoked. Reconnect from Settings → Integrations. **You work across more than one Jira site.** Tana connects one Jira site at a time, the one you authorize when you connect. To point it at a different site, reconnect from Settings → Integrations and choose the other site. **A status change didn't work.** The AI can only move an issue along the transitions your workflow allows from its current status. If the path isn't available, change the status in Jira directly. --- #### Connect Linear URL: https://tana.inc/learn/guides/connect-linear Set up the Linear integration so the AI can create and update issues from your meetings and conversations, and search issues to avoid duplicates. Tana connects to your Linear workspace so the AI can turn what you discuss into filed and updated issues, without you switching tools. Talk through a bug or a task in a meeting or in chat, and the issue gets created while the context is fresh. This guide walks through the one-time setup, what to expect from the authorization flow, and how to start using the integration in chat. ## What you can do with Linear in Tana Once Linear is connected, ask the AI in chat to: - **Create issues**: file a new issue in a team, with title, description, priority, labels, assignee, and project - **Update issues**: change fields on an existing issue, or move it to a new workflow state - **Search issues**: look up existing issues by keyword, useful for checking duplicates before you file a new one - **List teams**: see the teams you have access to, so issues land in the right place - **Attach meeting screenshots**: include screenshots shared on a call in the issue description, so it keeps its visual context You can also attach skills to Tana types that drive Linear actions one-click from your documents. See [Skills](/learn/features/skills) for how that works. ## Connect Linear You only need to do this once. The connection is per-account, not org-wide. 1. Open Settings with , or from the account menu in the sidebar. 2. Open **Integrations**. 3. Find **Linear** under personal integrations and click **Connect**. 4. A Linear authorization window opens. Sign in if you aren't already. 5. Review the access Linear lists for Tana, then approve it. 6. The window closes and Tana shows **Connected** next to Linear. ## Permissions Tana requests Tana asks Linear for the access it needs to work with your issues: reading your teams and issues, and creating or updating issues. Tana never reads or writes anything outside what you approve during connect. Your access token is stored securely on Tana's servers, never in your browser. You can review or revoke Tana's access at any time from your Linear account settings. ## Use it in chat Once Linear is connected, the AI can call Linear tools when a prompt asks for it. Some prompts to try: - "Check Linear for open issues about the login bug, then file a new one if there isn't a duplicate." - "Create a Linear issue on the Web team: write up the bug we just discussed and set priority to High." - "Move ENG-241 to In Review." - "What teams do I have access to in Linear?" The AI picks the right Linear tool from your prompt and asks for clarification when something is ambiguous, for example which team, or which of two similar issues. ## How Tana files an issue A few things happen automatically when the AI creates or updates an issue: - **Duplicate check**: when you ask it to, the AI searches existing issues first, so you don't file the same bug twice. - **Formatted descriptions**: the AI writes the description in markdown, which Linear renders natively. - **Screenshots as context**: if the issue comes from a meeting, the AI can embed screenshots shared on the call in the description. To change status, the AI moves the issue to another workflow state in its team. ## Disconnect Linear To remove the connection from Tana: 1. Open Settings → **Integrations**. 2. Open the **Connected** menu next to Linear and choose **Disconnect**. The same menu has **Reconnect**, which you can use to refresh access. Disconnecting stops the AI from reading or writing to Linear but does not change anything already in Linear or in Tana. To fully revoke Tana's access from Linear's side, also revoke it from your Linear account settings. ## Troubleshooting **The AI says it can't reach Linear.** The connection has likely expired or been revoked. Reconnect from Settings → Integrations. **You're not sure where an issue landed.** The AI returns a link to each issue it creates, so open that to jump straight to it in Linear. **A status change didn't work.** The AI can only move an issue to a workflow state that exists in its team. If the state you want isn't there, set it in Linear directly. **Something else?** See [Integrations](/learn/features/integrations) for the broader integration overview, or reach out at [support@tana.inc](mailto:support@tana.inc). --- #### Connect Pipedrive URL: https://tana.inc/learn/guides/connect-pipedrive Set up the Pipedrive integration so the AI can find contacts, companies, and deals, log notes to your CRM, and create or update records from your meetings and conversations. Tana connects to your Pipedrive account so the AI can pull CRM context into meeting prep and push outcomes back to Pipedrive, without you switching tools. This guide walks through the one-time setup, what to expect from the authorization flow, and how to start using the integration in chat. ## What you can do with Pipedrive in Tana Once Pipedrive is connected, ask the AI in chat to: - **Find people and companies**: search persons by name, email, or phone, and organizations by name - **Read a person's details**: pull a contact's full profile when you need more than the search result - **List deals**: see recent deals, optionally filtered by status such as open, won, or lost - **Log notes**: add a note to a person, an organization, or a deal - **Create deals**: open a new deal with a title, value, and currency, linked to a person or organization - **Update records**: change a person or organization's name, owner, labels, or custom fields, and link a person to an organization You can also attach skills to Tana types that drive Pipedrive actions one-click from your documents. See [Skills](/learn/features/skills) for how that works. ## Connect Pipedrive You only need to do this once. The connection is per-account, not org-wide. 1. Open Settings with , or from the account menu in the sidebar. 2. Open **Integrations**. 3. Find **Pipedrive** in the list and click **Connect**. 4. A Pipedrive authorization window opens. Sign in if you aren't already. 5. Review the access Pipedrive lists for Tana, then approve it. 6. The window closes and Tana shows **Connected** next to Pipedrive. ## Permissions Tana requests Tana asks Pipedrive for the access it needs to work with your CRM records: reading persons, organizations, and deals, and creating or updating persons, organizations, deals, and notes. Tana never reads or writes anything outside what you approve during connect. Your access token is stored securely on Tana's servers, never in your browser. You can review or revoke Tana's access at any time from your Pipedrive account settings. ## Use it in chat Once Pipedrive is connected, the AI can call Pipedrive tools when a prompt asks for it. Some prompts to try: - "What's the latest on the Acme deal? Find the contact and their open deals." - "Log this meeting as a note on Sarah Chen and on the Acme expansion deal." - "Create a deal called Acme expansion worth 48000 USD and link it to Acme Corp." - "Update Sarah Chen and link her to the Acme Corp organization." The AI picks the right Pipedrive tool from your prompt and asks for clarification when something is ambiguous, for example which Sarah, or which deal. ## Working with custom fields Pipedrive custom fields are identified by an internal field key, and their values come from how your Pipedrive account is configured. The AI can set custom fields when you name the field and the value, but for reliable updates, refer to them the way they are configured in Pipedrive. Updating labels or custom fields replaces the existing values on that record. ## Disconnect Pipedrive To remove the connection from Tana: 1. Open Settings → **Integrations**. 2. Click **Disconnect** next to Pipedrive. Disconnecting stops the AI from reading or writing to Pipedrive but does not change anything already in Pipedrive or in Tana. To fully revoke Tana's access from Pipedrive's side, also remove it from your installed apps in your Pipedrive account settings. ## Troubleshooting **The AI says it can't reach Pipedrive.** The connection has likely expired or been revoked. Reconnect from Settings → Integrations. **The AI created a record but I can't find it.** Records the AI creates are scoped to the Pipedrive account you connected. If you have access to more than one Pipedrive account, double-check you're looking at the right one. **A custom field update was rejected.** Pipedrive rejects updates when the value isn't valid for how the field is configured. Use the field key and the value as they are set up in your Pipedrive account. **Something else?** See [Integrations](/learn/features/integrations) for the broader integration overview, or reach out at [support@tana.inc](mailto:support@tana.inc). --- #### Connect Tana to Tana Outliner URL: https://tana.inc/learn/guides/connect-tana-outliner Connect your Tana Outliner graph to Tana with Remote MCP, so Tana's AI can search and read your nodes, pull them into documents and meetings, and write new content back into Outliner. If you use both Tana and Tana Outliner, you can connect them so everything in your Outliner graph is available to Tana's AI chat. Ask a question in Tana and it searches your Outliner nodes, follows the references around them, and writes what it finds into a Tana document or a meeting. It works the other way too: from a Tana chat you can create tagged nodes and todos exactly where they belong in your Outliner hierarchy. The connection runs over Tana Outliner's [hosted MCP server](https://outliner.tana.inc/learn/features/local-api-mcp#hosted-mcp-server-beta), its **Remote MCP**. Unlike the [local MCP server](https://outliner.tana.inc/learn/features/local-api-mcp), which runs inside the Outliner desktop app on localhost, the hosted server is reachable over the internet. There is nothing to keep running and no token to manage, and Tana can reach your graph with the desktop app closed. This guide covers the one-time setup, what the AI can do once connected, and the workflows worth building on top of it. ## Connect Tana Outliner Remote MCP is in beta and requires a paid Tana Outliner plan. The local MCP server stays free. You only need to do this once. The connection is per-account: each teammate connects their own, and the AI reaches the Outliner graph with your access. Open Settings with or from the account menu in the sidebar, then open **Custom MCP servers**. Click **Add server** and enter the Tana Outliner hosted URL: ``` https://app.tana.inc/mcp ``` Give it a display name you will recognize in chat, for example "Tana Outliner", and add it. Tana Outliner asks you to allow the connection. Click **Allow**. On the desktop app the authorization opens in your default browser, so passkeys and password managers work. Connecting adds a linked [skill](/learn/features/skills) for the server, named after the server with an "MCP:" prefix. Open the **Connected** menu on the server and choose **Configure** to see the tool catalog: search, read, and write tools from Outliner, with the server URL and when the list was last synced. You can allow or deny each tool individually, and **Refresh tool catalog** re-reads the list after Outliner ships new ones. That is the whole setup. Tana's AI now calls Outliner's tools like any other tool: ask something that needs your Outliner graph and it reaches for them on its own. You can also point at the skill explicitly with `/` when you want to be sure which one it uses. ## What the AI can do in Outliner Once connected, ask Tana's AI in chat to: - **Search your graph**: find nodes by text, supertag, field value, or date, the same conditions you would build in a live search - **Read and navigate**: read a node and its children, and walk the tree around it - **Work with supertags**: read a supertag's schema, apply tags, and set field values - **Create and edit nodes**: add nodes anywhere in the hierarchy, including today's day node, and check off todos A good first prompt is simply asking what it can do. Start a chat and ask it to list the Tana Outliner tools it has. It reports back what it can reach in your graph, which is the quickest way to confirm the connection is live. Each tool call shows in the chat as it runs, so you can watch it work: which skill it picked up, then the individual calls as it lists your workspaces, searches, and reads the nodes it finds. Ask in plain language. The AI picks the right Outliner tool from your prompt, and when a request needs several, it chains them. ## Prepare a meeting from your Outliner notes Years of thinking sit in your Outliner graph: daily notes, tagged people, project trees, reading notes. Tana is where the meeting happens. Connecting them means you stop retyping the first from memory into the second. Ask before the call, in a normal Tana chat: > "Look in Tana Outliner for everything I have written about Nadia, in my daily notes and anywhere she is referenced, and write me a prep doc in Tana with the open threads from our last few conversations." The AI follows references rather than reading only the node you named, so it picks up the mentions scattered across day nodes, not just the `#person` node itself. What comes back is a brief you can pin to the meeting, not a list of links. ## Look things up during the meeting Tana records the meeting as it happens, and the meeting's private chat can reach Outliner mid call. When something comes up that you know you wrote down two years ago, you can ask for it without leaving the call. > "What do my Tana Outliner notes say about the Hoffman pilot? Add a short summary to the meeting notes." See [Prepare for meetings](/learn/guides/prepare-for-meetings) for the rest of that workflow. ## Send the outcome back to Outliner The connection runs both ways, which matters if Outliner is still where you run your days. After a meeting in Tana, put the results where you will actually see them: > "Add the action items from this meeting to today's day node in Tana Outliner as todos, tagged `#task`." > "Create a node for the Hoffman pilot under my Projects node in Tana Outliner, tag it `#project`, and set Status to Active." Name the destination and the supertag. The AI writes into the hierarchy you point it at, and it can set fields as it goes, so what lands is a proper tagged node rather than a loose bullet you have to file later. ## Reuse a supertag as a Tana type If you are moving a workflow across rather than a single note, start from the schema you already built. The AI can read a supertag's fields in Outliner and build the matching [type](/learn/features/types) in Tana. > "Read the `#book` supertag schema in Tana Outliner and create a matching Book type in Tana with the same fields." That gets you a structure your Outliner content can land in, instead of a pile of untyped notes. ## Save a working flow as a skill Once a sequence works, tell Tana to turn it into a [skill](/learn/features/skills). It reads what you just did and writes a repeatable version, keeping the phrasing that worked. Run it by name from the slash menu afterwards. > "Turn what we just did into a skill called Prep from Outliner, so I can run it before any 1:1." Skills are also how you combine this with the rest of your stack. A skill can pull context from Outliner, check a related issue in [Linear](/learn/guides/connect-linear) or [GitHub](/learn/guides/connect-github), and file the result in Tana, all in one run. ## Run agents while your machine is off Because the server is hosted, anything that speaks MCP can reach your Outliner graph around the clock, not just Tana. The same URL works for Claude Desktop, Codex, and always-on cloud agents, with nothing running on your desktop. A small cloud agent reachable from your phone can query your graph while you are away from your computer, which is a real change for a graph that has always been heavy to load on mobile. ## Tips - **Name the destination when you write.** The AI does not guess where a node should land. Name the parent node, and the supertag if it should have one. - **Say Tana Outliner when you mean Tana Outliner.** Naming it in the prompt is the surest way to send the AI to your Outliner graph rather than to your Tana content. - **Ask for node IDs on anything you may want to update later.** With the ID in hand, the AI can go straight back to that exact node instead of searching for it again. - **Deny tools you do not want.** The per-tool controls in the skill panel are the cleanest way to keep writes out of a graph you would rather keep read-only. A denied tool is never offered to the AI, and it stays denied across a catalog refresh. ## Troubleshooting **Tana says it cannot reach Outliner.** The authorization has likely expired or been revoked. In Settings, Custom MCP servers, open the **Connected** menu on the server and choose **Reconnect** to authorize again. Your per-tool Allow and Deny settings are kept. **The AI answers from Tana instead of Outliner.** Say Tana Outliner in the prompt, or pick the Tana Outliner skill with `/`, so it knows which graph you mean. **The AI cannot find a node you know exists.** Name the supertag or the workspace it lives in, for example "nodes tagged `#project`", and it has something exact to search on. ## Disconnect In Settings, Custom MCP servers, open the **Connected** menu on the server and choose **Disconnect**. Its tools stop being available to the AI. Nothing already created in either tool changes. ## Go deeper For the full walkthrough, Mark's [Remote MCP systems lab](https://outliner.tana.inc/learn/tutorials/tana-systems-lab-remote-mcp) runs through the setup and the workflows above in detail, including building the skills and pointing a cloud agent at your graph. See also [Custom MCP servers](/learn/features/integrations#custom-mcp-servers) for how Tana connects to any MCP server, and [MCP](/learn/features/mcp) for the reverse direction, external tools reaching into Tana. --- #### Create AI agents with skills and integrations URL: https://tana.inc/learn/guides/agents-and-skills Create your own AI agents in Tana just by describing them: assistants you equip with skills and integrations that work on demand, in meetings, or on a schedule. {/* DRAFT for review. Doubles as the narration spine for a short explainer. */} You can create your own AI agents in Tana just by describing what you want. No setup, no prompt engineering. An agent is an assistant you delegate to. You equip it with **skills** (reusable instructions) and **integrations** (connected tools like GitHub, Slack, and Linear), and it works for you on demand, in a meeting, or on a schedule while you sleep. And because agents and skills are shared, building a good one is something your whole team gets, not just you. ## What an agent is An [agent](/learn/features/agents) is an assistant you configure for a job and then talk to. It has its own prompt, its capabilities, an optional voice, and an optional [schedule](/learn/features/agents#schedules). Converse with it in chat or by voice, or set it to run on its own. Keep an agent to yourself, share it with a teammate, or open it to the whole org. A capability one person builds becomes the team's with nothing extra to set up, and a useful agent becomes a shared teammate. ## What you equip it with An agent's [capabilities](/learn/features/agents#agent-capabilities) decide what it can do, in two parts: - **Skills** are reusable instructions for specific jobs: "file this as a GitHub issue," "summarize this meeting in our format," "draft a PRD from these notes." A skill does the same job the same way every time, for everyone who uses it, so your whole team files bugs or drafts PRDs in one consistent shape. You can also run a skill on its own, with `/` in [chat](/learn/features/chat), from an item's **Send to** menu, or from the Skills button in a [meeting](/learn/features/meetings#during-the-call). - **Integrations** connect the agent to your tools, like GitHub, Slack, and Linear, so it can act in them rather than just talk about them. You toggle both on in the agent's Customize panel; together they set what the agent can read, create, and reach. ## Build a meeting-prep agent You do not configure agents by hand, you describe them. Here is the whole flow, building one agent that preps you before every meeting, the job [Walk into every meeting prepared](/learn/guides/prepare-for-meetings) puts on autopilot. Click the pencil button at the top of the sidebar, or ask in [chat](/learn/features/chat): "Create an agent that preps me for every meeting by pulling context on the people I'm meeting with and what we discussed last time." Tana scaffolds the agent from that one description: its prompt, a name, and a starting set of capabilities. Open the agent's Customize panel and give it what the job needs. Turn on a [skill](/learn/features/skills) that writes the prep in your format ("summarize our last meeting and list open items per attendee"), and an [integration](/learn/features/integrations) when the context lives in another tool, like Slack or your CRM. Together, skills and integrations set what the agent can read, create, and reach. Not quite right? Tell it: "also include any open tasks assigned to the people in the meeting." You refine an agent the same way you created it, in plain language, never by editing settings by hand. See [Agents](/learn/features/agents#creating-an-agent) for the full reference. Now choose when it runs: - **Chat or voice**, when you want to ask it for a prep on demand. - **On a [schedule](/learn/features/agents#schedules)**, so it runs while you sleep and tomorrow's preps are waiting in your Library each morning. - **As a space's default assistant**, picking up new chats started in your team's space. Because the agent is a doc, you can keep this one to yourself or share it so the whole team gets the same prep. The more of your repeatable work you hand to agents, the more Tana does on its own, so you spend your time on the conversation, not the busywork. --- #### Get outcomes to the right people and tools straight from the meeting URL: https://tana.inc/learn/guides/get-meeting-outcomes-to-people-and-tools An outcome is finished work. While the meeting runs, ask the AI to update your proposal with what the customer just said, file their feature request in Linear, Jira, or GitHub, send them the updated proposal, post the summary to Slack, and pin what is still open before you leave. {/* DRAFT for review. Teaching exemplar: the lesson is that the Share button and Send to menu are the same everywhere, so learning them on an outcome teaches them for everything. Growth framing per VP1: the work completes in the meeting, and an outcome is finished work, never "just a document". One running scenario (a customer meeting about a pilot proposal) matches the screenshots. Five parallel moves, so H2 per move rather than a Steps sequence. Sequel to prepare-for-meetings (before the meeting) and sibling of catch-up-and-act (chat as retrieve-then-act). */} You are twenty minutes into a customer meeting. They have just told you the pilot needs to cover twice as many learners as your proposal assumes, and asked whether companies can upload their own documentation. In most tools that is two lines in someone's notes and an hour of work tonight. In Tana it is work you finish before the meeting ends. [Capture](/learn/features/meetings#during-the-call) turns what was just said into outcomes while the conversation continues, and this guide covers the five moves that get those outcomes where they belong: into your tracker, in front of the room, out to the customer, into the channel your team reads. Each one uses the same **Share** button and [Send to](/learn/features/docs#send-to) menu that every item in Tana has, so learning them on a meeting outcome teaches them everywhere. And nothing here has to wait for the meeting to end. ## Before you start You need a meeting with at least one accepted outcome. Anything the AI captures waits as a [proposal](/learn/features/proposals) until someone accepts it, so work through the [review list](/learn/features/meetings#review-and-use) and keep what is right. Two of the moves need a connection first, both from Settings: - **Slack**, connected to your personal account, not only the organization-level app your admin installs - **The tracker you file work in**, [Linear](/learn/features/integrations#linear), [Jira](/learn/features/integrations#jira), or [GitHub](/learn/features/integrations#github), plus any [coding tools](/learn/features/integrations#coding-tools) you want the AI to launch Sharing and pinning need nothing set up at all. ## Ask for the outcome the meeting actually needs The outcomes Tana captures on its own are the common ones. When you need something specific, ask for it in the meeting's own [Private Meeting Agent](/learn/features/meetings#chat-in-the-call), a chat only you can see, so you are not interrupting the room to get work done: - _"Update the proposal based on what the customer said about learner numbers."_ - _"Capture the documentation upload request as a feature request."_ - _"Draft the follow-up email to send them tonight."_ - _"Log the decision to run the pilot in two phases, and who agreed to it."_ The first one is the one to notice. The AI does not only create new outcomes from the meeting, it updates work you already have, so the proposal reflects what the customer said minutes after they said it. Changes come back as [proposals](/learn/features/proposals) you accept, and asking again refines the existing outcome rather than leaving you with two. For more of what is worth asking for, browse the [outcomes Tana can produce](/outcomes). When a guest is in the meeting, AI is switched off in the shared Chat With Everyone to keep your organization's data out of a chat the guest can read. Your Private Meeting Agent keeps working, which is why it is the right surface for customer meetings. ## File the work in your tracker before the meeting ends That feature request is worth more as an issue in the tool your team actually works from. Open [Send to](/learn/features/docs#send-to) on the outcome and pick the skill or integration for [Linear](/learn/features/integrations#linear), [Jira](/learn/features/integrations#jira), or [GitHub](/learn/features/integrations#github). Filing from a meeting beats filing by hand, and not by a little. The AI reads the transcript for the detail that never survives a summary, pulls in the screenshots from anyone's screen share, and asks which repository or project when the source material does not make it obvious. Once the issue exists, the chat links straight to it, so you can say "it's filed, here it is" while the customer is still in the meeting. For work that goes to code, hand the whole context to a [coding tool](/learn/features/integrations#coding-tools) as a self-contained prompt. Ask in the words you would use anyway, _"open a pull request for this"_, and the AI assembles the prompt and launches the tool, leaving an **Open in** button in the conversation. Claude Code runs in a project folder you configure in Settings, Coding Tools; Codex, Cursor, GitHub Copilot, and Lovable are there too, and the AI asks which one when several are enabled. Claude Code needs the desktop app on macOS, because it launches a local CLI in a folder on your machine. On the web app the button is disabled with a "Desktop app required" note, and on Windows it reads "Not supported on Windows yet." ## Put the outcome in front of the room, then beyond it Anything captured in a meeting is visible to the participants by default, and the capture popover tells you who that is before you commit. To make an outcome part of the meeting itself, use **Pin to meeting** from its menu: it lands on the meeting surface where everyone, guests included, can open it and see the same thing you are looking at. Tiles carry a state control and an assignee, so you can move the feature request to In Progress or hand it to a teammate without leaving the meeting. Beyond the room, each outcome has its own **Share** button, and access is managed [per outcome](/learn/features/spaces#sharing-outcomes), independently of the meeting it came from. You can send the customer their updated proposal without giving them the meeting, the transcript, or anything else that was discussed. Add a teammate by name, or type any email address for someone outside your organization. For a reader who should not have to sign in at all, turn on **Anyone with the link can view** and send the URL. **Copy Public Link** does both steps at once, and **Download Markdown** gives you the file when the outcome needs to travel somewhere else entirely. You will know it worked when a public link opens the read-only [public viewer](/learn/features/spaces#link-sharing) in a browser where you are not signed in. ## Post the update to Slack by asking The teammates who were not in the meeting read updates where they already are, which is a channel. Rather than copying anything across, ask in the same chat: _"Share a summary of this meeting to the #gtm channel in Slack."_ With your Slack account connected, the AI can list channels and post to a channel or a direct message on your behalf, and it confirms in the chat once the message is sent. You named the audience; it wrote for that audience. No reformatting, and no deciding by hand which parts matter. The traffic runs the other way too. Mention `@tana` in any Slack thread and Tana joins the conversation with access to what you are allowed to see, so a colleague who replies with a question gets an answer without opening Tana. See [Slack](/learn/features/integrations#slack) for the full setup. ## Pin what is still open before you leave The ticket is filed, the proposal is updated, and the channel knows. What is left is the work you are carrying, and two pins keep it in front of you without a to-do list to maintain. **Pin to sidebar**, from an item's right-click or toolbar menu, puts it in your own sidebar, either at the top level or filed into a section you name. An item sits in one place in that tree, so pinning it somewhere else moves it rather than duplicating it. **Pin to date** puts it under Pinned on a specific day, on [Home](/learn/features/home#pinning-items-to-dates) and the [Calendar](/learn/features/calendar), which is the right choice for anything with a deadline: pin the proposal to Thursday and it is waiting for you on Thursday. Use the sidebar for what you are carrying now, and dates for what is coming due. Everything else stays findable in its space without a pin at all. ## If something does not work - **The AI is missing from the shared chat.** A guest is in the meeting. AI stays off in Chat With Everyone for the rest of the meeting, even after they leave. Use your Private Meeting Agent instead. - **The AI says it cannot post to Slack.** The organization-level Slack app gives your team the `/tana` command; posting from chat needs your own account connected under Settings, Integrations. - **A public link you already sent stops working.** Turning link sharing off applies immediately, and the old URL shows a sign-in prompt instead of the document. Turn sharing back on and the same URL resolves again. - **The coding-tool button says the prompt is too long.** A note appears above it with **Copy full prompt**, which puts the complete prompt on your clipboard. The shortened link still works as a secondary action. - **Filing an issue asks you a question first.** That is deliberate. When the source material does not name a repository or project, the AI asks rather than guessing. ## Every meeting adds to what the next one knows The work is done, and something else has quietly happened: the decision, the issue, and the proposal are all connected to the meeting they came from. In six months you can still find why the pilot covers twice as many learners, and who agreed to it. That is what makes the next meeting with this customer different. You walk in with the history rather than reconstructing it, whether you [prepare for it yourself](/learn/guides/prepare-for-meetings) or leave that to an agent overnight. And after a few days away, the same record is what lets you [catch up and act in one pass](/learn/guides/catch-up-and-act). --- #### Getting started with Tana URL: https://tana.inc/learn/guides/getting-started New to Tana? The fastest way to see what it does is to run a meeting with a colleague and watch the conversation turn into finished work your team shares, no setup needed. {/* DRAFT for review. Front door for a brand-new, empty-workspace user. Cold-start aware (the conversation is the input; chat creates from scratch, it does not query data you do not have yet). Drives the key activation move: invite a colleague and collaborate. */} Tana is an agentic meeting platform built for teams. You hold your meetings in it, and as you talk the AI turns the conversation into finished work, a filed bug, a logged decision, a drafted recap, that you review before anything lands. You are starting with a clean slate, and that is fine: you do not need to fill it first, because the conversation is the input. The single best way to see what Tana does, and the move that matters most, is to run a meeting with a colleague. ## Run your first meeting, with someone You need no setup and no existing notes. From [Home](/learn/features/home), open the create menu (the pencil icon in the sidebar) and choose **Meeting** to get a call link. **Invite a teammate** from your organization, or send the guest link to anyone outside it, they join with just a name, no account. Then click **Join Meeting**. Have a real conversation, a planning chat, a bug triage, a quick design look. You are not taking notes. When something concrete comes up, hit the **+** in the call toolbar, pick how far back to look, and choose what it is, a bug, a decision, a task, and Tana drafts it from that part of the talk. When the call ends, it writes the summary and pulls out anything you missed. You review each outcome, edit what is off, and accept what you want. Because the meeting is shared, everyone in it has that work right away. That is the whole point: a conversation becomes finished work your team can act on, with no write-up. ## Try it on your own first Want a quick taste before you pull anyone in? Start a [chat](/learn/features/chat) from Today and ask the AI to make something from scratch: - "Draft a one-page PRD for a referral feature." - "Turn these rough notes into a project plan." (paste the notes) - "Outline a technical spec for a rate limiter." Everything it produces is yours to review and keep. Chat gets far more powerful as you go, once you have meetings and have connected tools like [GitHub](/learn/guides/connect-github) or [Linear](/learn/guides/connect-linear), it draws on everything your team has done and acts in the tools you already use. ## What is worth knowing early - **The AI [proposes](/learn/features/proposals), you decide.** Everything you would act on, outcomes, docs, edits, lands as a proposal you review and approve first. The one thing it writes for you automatically is the meeting summary. That control is what makes it safe to let the AI do real work. - **Everything you create stays in your [Library](/learn/guides/organize-your-context)**, findable from the start. No setup, and nothing gets lost. - **Add structure only when it earns its keep.** Once you are making many of a kind, bugs, clients, decisions, give them a [type](/learn/guides/organize-your-context) so they share fields and land on a board. When a body of work needs its own people, put it in a space, and where something lives is also who can see it. - **It compounds.** Every meeting and chat feeds your shared context, so a few weeks in you can [ask](/learn/guides/catch-up-and-act) "what did we decide about this, and why" and get an answer grounded in your team's actual work. ## Get your team in This is the step that changes everything. Tana is built for working together, and almost nothing about it is as good alone: shared meetings, shared outcomes, a context that the whole team builds and draws on. Invite the people you actually work with, and the next meeting you run together already produces work everyone has, while the AI gets sharper about your decisions, your roadmap, and your code with every conversation. When you are ready for the daily rhythm, [How to work in Tana](/learn/guides/how-to-work-in-tana) goes deeper. --- #### How to work with your team in Tana URL: https://tana.inc/learn/guides/how-to-work-in-tana The recommended way for a product or tech team to work in Tana together: how meetings turn into finished work, where to capture, when to add spaces and types, and what a productive day looks like. {/* DRAFT for review. Orientation cornerstone for the Guides section; the other guides link back to this one. Written for the launch ICP (product and tech teams). */} Tana is an agentic meeting platform: your team meets in it, and the AI turns those conversations into finished work, bugs filed, PRDs drafted, decisions logged, before the call ends. It is collaborative from the first meeting. The work the AI produces is shared, so teammates pick it up without a handoff, and the context builds for everyone. This guide is the recommended way to work in it together: how the loop runs, how to set up, and what a productive day looks like. You do not need to set much up to start. Read this once for the shape, then let structure grow as your work earns it. ## The core loop: meetings in, finished work out One loop sits at the center of Tana, and everything else serves it. 1. **You meet.** Hold the call in Tana, or [capture an external Zoom, Teams, or Meet call](/learn/guides/capture-external-meetings) from the desktop app. Tana transcribes as you talk and reads any screen you share. 2. **The AI does the work, live.** As the conversation happens, it pulls out the real outputs, a bug with steps to reproduce, a decision with its rationale, a drafted PRD, and can file them straight into Linear, GitHub, or your other tools. 3. **You approve.** Nothing lands on its own. Every change arrives as a [proposal](/learn/features/proposals) you review, edit, and accept. 4. **Outcomes scatter.** Once accepted, tasks show up in your [Tasks](/learn/features/tasks) view, docs and artifacts in the right space, issues in your tracker, ready for whoever needs them. The shift to feel: you are not taking notes to process later. The work takes shape during the conversation, and you leave the call with it mostly done. Two things make this safe to rely on. The **proposal gate** means the AI can draft, file, and update, but you see every change before it happens, so nothing silently lands in your team's space. And **what's captured** is only the transcript and the screens you share, never anyone's camera or face. For sensitive moments, [off the record](/learn/features/meetings#off-the-record) pauses all capture for everyone. ## What you're working with A few building blocks, in roughly the order you meet them: - **[Meetings](/learn/features/meetings)** are the main way work gets in: video, transcription, screen capture, and live capture of outcomes. - **[Proposals](/learn/features/proposals)** are the review gate: every AI change, seen before it lands. - **The [Library](/learn/features/interface)** is where everything your team creates lives and stays findable, even before you add any structure. - **Spaces and types** are how you add structure once it earns its keep: a [space](/learn/features/spaces) is where a team works together, with its own members and shared context; a [type](/learn/features/types) is for a kind of thing you have many of, like Bug or Decision. - **[Agents](/learn/features/agents) and [skills](/learn/features/skills)** are how you make the AI yours: reusable instructions and assistants that do recurring work your way. Meeting and Task are built in. Bug, Decision, and the rest are types you define, so they carry the fields your team cares about. For how spaces and types fit together, see [Manage your context with spaces and types](/learn/guides/organize-your-context). ## Set yourself up You can do real work on day one with almost no setup. The light path: 1. **Connect your calendar** so your meetings show up on [Home](/learn/features/home) and the [Calendar](/learn/features/calendar). 2. **Capture into the Library to start.** Quick notes, chats, and meeting outcomes all land there and stay searchable. No structure required. 3. **Add a type when a kind repeats.** Once you have filed a few bugs or logged a few decisions by hand, make a Bug or Decision type so they share fields and the AI files new ones in the same shape. 4. **Add a space when work needs its own people.** A team, or a major project: a space carries its own members, types, and AI instructions. The rule of thumb: let structure follow the work, not precede it. ## A day in Tana Here is the loop across a real day on a product or tech team. **Before a meeting.** Open the Today view to see what's coming, then ask the AI to prep you: "pull what we decided with Acme last month and any open bugs they reported." An [agent](/learn/features/agents) can do this on a [schedule](/learn/features/agents#schedules), so the prep is waiting each morning. **During standup.** A bug comes up. Capture it as you talk, and Tana files it in Linear with steps to reproduce and a screenshot from the shared screen, before standup ends. No after-the-fact triage, and the whole team sees it land. **In a customer interview.** Tana extracts the pain points, drafts a customer-journey [artifact](/learn/features/artifacts), and starts a PRD from what was actually said. You share it with the team minutes after the call. **In a design review.** Feedback lands on the right issues, updated in place with the screenshot of what was on screen, so the designer leaves with the work tracked rather than a pile of notes to reconstruct. **Right after.** Review the proposals the meeting produced, the summary, the outcomes, the issues. Edit anything that needs it, then accept. This is the one habit worth keeping: review before you move on. **End of day.** Check your [Tasks](/learn/features/tasks) view for what you and the team committed to, and pin anything urgent to tomorrow. The payoff to notice: every one of these leaves a real artifact, in the tool your team already uses, with no separate write-up step. ## Working as a team Tana is most valuable with your team in it, and getting them there is low-friction. **Adoption is joining a call.** When you run a meeting in Tana, everyone in it is working with the AI, with no setup or training. That is the simplest way to get a team using AI together, instead of each person stitching together their own stack of note-takers, bots, and tools. **Share outcomes, not just meetings.** A decision or task can go to people who were not in the room, so the right colleagues see the result without the whole transcript. Where something lives decides who can see it, so a private space stays private and shared work reaches the team. See [Spaces and sharing](/learn/features/spaces#access-modes). **Context compounds.** Every meeting feeds a shared memory. After a few weeks, "what did we decide about auth, and why?" has an answer grounded in the actual calls. The more your team works in Tana, the sharper the AI gets about your codebase, your roadmap, and your decisions. --- #### Manage your context with spaces and types URL: https://tana.inc/learn/guides/organize-your-context How items, spaces, and types fit together in Tana, when to reach for each, and how where something lives decides who on your team can see and work on it. {/* DRAFT for review. Doubles as the narration spine for a walkthrough. The "Spaces vs types" model below reflects Theo's guidance (type = a kind you have many of; space = a few broad collaborative contexts; relate via link fields) and is the key thing to confirm before publishing. */} You can start using Tana without setting anything up. Much of your context is born in [meetings](/learn/features/meetings) and [chats](/learn/features/chat), the tasks, decisions, and notes a conversation leaves behind. All of it lands in your **Library**, always within reach. You add structure, spaces and types, only once it earns its keep. It earns it three ways: - it **shapes** the work, with shared fields and relationships - it makes the work **easy to find** later, for you, your team, and the AI - it sets **who can see and act** on it That last one is in play from day one. Tana is built for working together. This guide walks through how items, spaces, and types fit together, and closes with a simple way to decide what to add and when. ## The Library holds everything The **[Library](/learn/features/interface#more)** is the single view of everything in your organization. It opens as a grid of **space cards**, with one **filter bar** (shared with [Search](/learn/features/interface#search)) to narrow things down. Anything that does not live in a space or on a meeting lives here. So before anyone makes a space or type, the Library already holds everything you and your team create, and nothing gets lost. This is the foundation: spaces and types are not separate places you move things into. They are organizing layers that show up _within_ the Library. The filter bar's round pills scope to a space, jump straight to meetings or chats, filter by who created an item, or open the type palette for any kind, your own types included. Think of the Library as the map your whole team shares, and spaces and types as structure drawn onto it. ## Spaces vs types: when to use which Two ways to add structure, for two different jobs. Use a **[type](/learn/features/types)** for a _kind of thing you have a lot of_: clients, bugs, candidates, invoices. A type gives every item of that kind the same fields, and lets you connect items to each other with link fields. Reach for one whenever you will create many of something and want to filter, track, or relate them. A Client, for instance, can carry fields like industry, owner, and renewal date, and your other items can link to it. Use a **[space](/learn/features/spaces)** for a _broad area of work that has its own people and context_: a team, a major project, a domain you work in. You will have a handful of spaces, not hundreds. Each one carries its own members, its own types and skills, and its own AI instructions. Spaces can nest. A space can hold [sub-spaces](/learn/features/spaces#nested-spaces) for real sub-areas with their own people, like a department space with a sub-space per team. But resist a Clients space with a sub-space per client: you would drown in spaces as your business takes off. Keep one space for the team that works with clients, make Client a type, and link your items to the right client. (With only a handful of clients total, a space each can be fine. Scale decides.) And you do not always need either. Plenty of work is fine living in your Library with no type and no space, until a repeating pattern, or a person you want to share with, makes the structure worth adding. ## What a space gives you A space is where collaboration happens. Adding someone to a space gives them access to everything in it, so a space is as much about _who works on this_ as _what goes in it_. On top of shared access, a space gives that work: - its **members**, and the access that flows from them - its own **[types](/learn/features/types)** and **[skills](/learn/features/skills)** - **AI instructions** that shape how the AI behaves for that work A chat or meeting inside a space follows that space's AI instructions and draws on its types and content, so the AI works the way that team needs. Access flows downward: from the organization, to a space, to an item. Deciding where something lives is also deciding who can see it. See [Spaces and sharing](/learn/features/spaces#access-modes). ## Where things land, and who can see them Whenever you create or capture an item, in a chat, in a meeting, or directly, it lives somewhere: in your Library, or inside a space. Where it lives sets two things at once: - **What it is:** give it a [type](/learn/features/types) to add structure, or leave it plain. - **Who can see it:** an item inherits the access of where it lives. Something in a private space is visible only to that space's members; something in your Library, at the organization level, is visible to everyone in the org. You can set this when you create the item, or move it later from its [location](/learn/features/docs#location-and-access) controls. When the AI proposes a new item, it shows where it will live before you accept, so you can redirect it into the right space. ## Retrieving context Pulling context back when you need it is the other half of managing your context. It works three ways, and spaces and types are what make them sharp: you can navigate to your work, ask the AI to surface it, or let what your team is doing come to you. **Navigate to it.** Start with the **Library**: type a query, or filter by space, by who created an item, or by type. Two focused views build on the same structure: - A **space home** is the dashboard for one space your team shares: its pinned items, a card for each [type](/learn/features/types#the-type-page) in it, and its content list. - A **[type page](/learn/features/types#the-type-page)** lists every item of a type, as a list or as a [kanban board](/learn/features/types#the-kanban-board) when the type has a workflow, so you get "all open Bugs" across the team in one place. **Ask the AI to surface it.** Because your work carries types and lives in spaces, the AI can pull it up the way you would describe it: "show me open bugs for Acme," "what did we decide in last week's planning meeting." A [chat](/learn/features/chat) inside a space also follows that space's AI instructions and draws on its content, so its answers stay grounded in the right context. **Let it come to you.** [Work](/learn/features/interface#work) is the board for what you owe people, and its **Notifications** card gathers what teammates did that involves you: assigned you a task, sent you a chat message, or are waiting in a call. **Recents** on [Home](/learn/features/home#the-cards) brings back what you were last in. Click any entry to jump to the source, so the work that needs you finds you. Spaces and types are layers within the Library, so anything you structure stays within reach there too. You never choose between structure and retrieval. Structure just gives you sharper ways to slice the same map, by hand or by asking. ## A simple way to start When you are not sure what to add, let the work tell you: 1. **Start in your Library.** Capture and create freely, with no type and no space. There is nothing to set up, and nothing gets lost. 2. **Add a type when you have many of one kind.** The moment you are creating clients, bugs, or candidates and want to filter, track, or relate them, give them a [type](/learn/features/types) so they share fields and connect to each other. 3. **Add a space when work needs its own people and context.** A team, a major project, a domain you work in: anything that wants its own members, skills, and AI instructions. Keep spaces few. 4. **Relate items with link fields, not more spaces.** When two kinds of work connect, link them with a field rather than spinning up a space for each. A Bug can carry a field pointing to the Feature it affects, so you can move between them without either needing its own space. You can always add structure later, and move an item to a new home once its shape is clear. Structure is something you grow into, not a decision you make up front. --- #### Tana Tutorials: The Basics of Tana URL: https://tana.inc/learn/guides/tana-learn-live-basics A recorded walkthrough of the basics in Tana and how they fit together: run a meeting, review the proposals it drafts, send the work to your tools, and organize everything with spaces and types so it stays findable. A guided walkthrough of the basics in Tana, the agentic meeting platform, and how the pieces fit together. You run a meeting, watch the AI draft proposals from what was said and shown, review and accept the ones you want, send that work to the tools you already use, and organize everything with spaces and types so your team can find it later. It is the fastest way to see the whole loop end to end. ## What Tana is [[0:00](https://youtu.be/lgZhRrXPf0M?t=0)] Tana brings AI and agents into the meeting itself, so you complete work as you talk instead of leaving with notes to process later. It sees both what is said and what is shared on screen, and closes the gap between the conversation and execution: you come out with work that is more or less ready to go, shipped to where work actually happens. See [Getting started](/learn/guides/getting-started) for the same idea in writing. ## Spaces [[3:14](https://youtu.be/lgZhRrXPf0M?t=194)] A [space](/learn/features/spaces) is a container for related work, like a folder or a project, with its own access control: everyone in the org, just you, or a named few. Everything you put in it inherits those permissions. You can also `@`-mention a whole space in a [chat](/learn/features/chat) to use it as reusable context, which is what sets Tana apart from a standalone AI tool. ## Types [[4:22](https://youtu.be/lgZhRrXPf0M?t=262)] A [type](/learn/features/types) (the equivalent of a supertag in Tana Outliner) says what something is: a bug, a dev task, a person. That helps the AI treat it correctly, gives it a template of fields like severity, and lets you link types together, for example an issue linked to a product area. [Track issues with a type](/learn/guides/track-issues-with-a-type) goes deeper. ## Inside a finished meeting [[7:28](https://youtu.be/lgZhRrXPf0M?t=448)] The walkthrough opens just after a meeting, in the after-meeting view: a digest with a title and description, the full raw transcript behind it, and the sections Tana draws out live while you speak. Everything here is collaborative with the people in your org from the start. ## The summary, and sharing [[8:31](https://youtu.be/lgZhRrXPf0M?t=511)] Tana writes a rich [summary](/learn/features/meetings) automatically. Every item carries its own access control, not just spaces, so you can hand a single summary to a customer with a guest link, or keep it to selected people, without changing anything else. ## Wrap-up and proposals [[10:09](https://youtu.be/lgZhRrXPf0M?t=609)] When the meeting ends, the wrap-up writes the summary, makes sure the digest is accurate, and drafts [proposals](/learn/features/proposals) for the actions that came up. In the demo that includes updating an existing issue with new customer-discovery context. You review each one and decide what lands, so everyone with access gets the richer context at once. ## Send work to your tools [[11:17](https://youtu.be/lgZhRrXPf0M?t=677)] From a proposal you can send the work, with its context, to Linear, Jira, or GitHub, or hand it to a coding agent like Claude Code that spins up a session and starts on the issue. You set these up once in Settings, then Integrations. See [Connect GitHub](/learn/guides/connect-github), [Connect Linear](/learn/guides/connect-linear), and [Integrations](/learn/features/integrations). ## Working during the meeting [[13:22](https://youtu.be/lgZhRrXPf0M?t=802)] Live, you have your call tiles, the live digest, and pinned talking points. The `+` menu is where you capture as you go: search for what you discussed and capture it from the last couple of minutes. Context is not just the transcript, if someone is sharing their screen, Tana takes that into account too, so the outcomes are richer. ## Every proposal has a type [[15:35](https://youtu.be/lgZhRrXPf0M?t=935)] Each proposal is labeled with its [type](/learn/features/types), issue, feature request, company, deal, and the AI considers what kind of thing it is while extracting. Because types carry fields and can have a skill attached (say, one that always sends issues to GitHub), this is what makes more powerful workflows possible. ## The Library [[16:50](https://youtu.be/lgZhRrXPf0M?t=1010)] Everything you create in Tana lives in the [Library](/learn/guides/organize-your-context): documents, tasks, artifacts, types, chats. Spaces are containers inside it, and they are where access control lives, the product space shared with the whole org, a research space limited to a few people. ## Browse and filter [[19:05](https://youtu.be/lgZhRrXPf0M?t=1145)] The Library is built for browsing. Filters let you see just meetings (today, upcoming, shared, or with a specific person), or switch to chats, canvases, agents, skills, artifacts, spaces, and types. [Organize your context](/learn/guides/organize-your-context) covers the model. ## Today, and the chat bar [[21:00](https://youtu.be/lgZhRrXPf0M?t=1260)] [Home](/learn/features/home) shows the meeting you are heading into and what came out of recent ones. The [chat](/learn/features/chat) AI has the same tools you do, plus current knowledge of Tana itself, so it can both do the work and explain how the app works. ## From a feature request to a PRD [[21:55](https://youtu.be/lgZhRrXPf0M?t=1315)] The session ends by running a [skill](/learn/features/skills), reusable instructions you save and rerun, to turn a captured customer feature request into a full product spec, by voice. Because the feature request holds the customer's own words and quotes, the spec links back to the source and flags open questions, shortening the path from a customer conversation to built product. --- #### Use Tana with Claude URL: https://tana.inc/learn/guides/use-tana-with-claude Connect Tana to Claude so you can pull meetings, tasks, and team knowledge into a Claude conversation as context, and save new docs and tasks back to Tana, with every write reviewed as a proposal first. Connect Tana to Claude so you can use your team's meetings, tasks, decisions, and documents as context in a Claude conversation, then save new work back into Tana when you're done. Tana connects to Claude through the Model Context Protocol (MCP), an open standard that lets Claude read from and write to external tools. Once connected, Claude can pull documents, meeting notes, and structured data out of Tana for context, and propose new content back into Tana. This guide covers the one-time setup, what Claude can access, and how to use the connection. ## What you can do with Tana in Claude Once Tana is connected, ask Claude to: **Pull context out of Tana** - **Get meeting context**: read transcripts, screen-share screenshots, attendees, and attached artifacts from a meeting - **Work with tasks**: list open tasks, or plan work from an existing task - **Search Tana**: full-text and semantic search across spaces for notes, docs, and decisions - **Read a specific document**: pull up a design doc, spec, or typed entry with all its fields - **Check calendar events**: see what's on today, or prep an upcoming meeting from its details **Save work back to Tana** - **Create documents**: save a summary, spec, or recap as a doc in Tana - **Create typed entries**: file a Bug, log a Decision, or add any typed document with the right fields - **Create and update tasks**: add follow-ups, or break an existing task into sub-tasks - **Update existing documents**: append notes or edits to a doc already in Tana Every change Claude makes is a **proposal** you review and approve in Tana. See [Working with proposals](#working-with-proposals) below. Nothing is written without your confirmation. ## Connect Tana to Claude You only need to do this once. The connection is per-account. Tana isn't in Claude's built-in connector directory yet, so you add it as a custom connector using Tana's MCP server URL: `https://home.tana.inc/mcp`. In Tana, **Settings → MCP** has a one-click copy for the URL and ready-made setup snippets. ### In Claude Code Add Tana with a single command: ``` claude mcp add tana --transport http https://home.tana.inc/mcp ``` The first time you use it, Claude Code walks you through signing in to Tana. ### In Claude Desktop or claude.ai Claude connects to Tana through Anthropic's connectors flow: 1. In Claude, open **Settings**, then **Connectors**. 2. Choose **Add custom connector**. 3. Paste the Tana MCP server URL: `https://home.tana.inc/mcp`. 4. Click **Connect**. A Tana sign-in window opens. Sign in if you aren't already, and grant access to the organization you want to connect. 5. The window closes and Tana shows as a connected tool in Claude. {/* TODO screenshot: Claude's "Add custom connector" dialog with the Tana MCP URL entered. */} Custom connectors are available on all Claude plans; free accounts can keep one at a time. For the latest steps on Claude's side, see Anthropic's guide to [custom connectors](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp). You can now reference Tana in any Claude conversation. ## What Claude can access Claude works only with the content in the organization you connected, and only when a prompt asks for it: - **Search**: full-text and semantic search across your spaces - **Read**: notes, typed entries and their fields, tables, backlinks - **Meetings**: transcripts, screen-share screenshots, event details, attendees, artifacts - **Files**: text-based attachments like markdown, CSV, and plain text, plus images; other files come back as metadata such as name, type, and size - **Create**: documents, tasks, and typed entries - **Update**: text, fields, task state, assignments, tables - **Delete**: remove documents, with your approval The Tana connector reads and writes only your Tana content. It does not access your Claude memory, chat history, or files you've uploaded to Claude. Tana captures the shared screen during a meeting, not participants' camera video, so the screen-share screenshots Claude can read never include people's faces. ## Using Tana in Claude Ask in plain language; Claude picks the right Tana tool from your prompt. Some prompts to try: - "Read the transcript from yesterday's standup in Tana." - "Find my notes about the Q3 roadmap in Tana." - "List all open bugs in Tana" or "Show me all Feature Requests assigned to me." - "Prep my 2 pm meeting: read the details and attached artifacts." - "Save this summary as a doc in Tana called 'API Migration Plan'." - "Add a task in Tana to review the pull request by Friday." ### Example workflows **1. Turn meeting outcomes into a draft** > "Read the transcript and outcomes from yesterday's design sync in Tana, then draft a one-page recap with agreed decisions and owners I can paste into Slack." Claude finds the event, reads the transcript and any pinned artifacts, and produces a recap grounded in what was actually said. It's read-only, so nothing is written to Tana. **2. Research across Tana and draft a spec** > "Search Tana for everything related to our authentication system, the bugs, design docs, and decisions, then draft a technical spec for the new SSO feature, citing the source docs." Claude searches across your spaces, reads each match, and produces a spec that links back to the source documents in Tana. Useful when context is scattered across many notes and meetings. **3. Capture work output back in Tana** > "I just finished this Q2 metrics analysis. Create a doc in Tana titled 'Q2 Metrics Analysis' with the content above, and create three follow-up tasks for the action items." Claude proposes the document and three tasks. They appear in Tana for review. You approve them and they land in Tana. ## Working with proposals Every write Claude makes, whether creating a document, updating a field, or deleting an item, goes through Tana's proposal system. The change appears as a pending proposal in Tana that you review and approve before it takes effect. This is the same review loop Tana's own Chat uses, so nothing is added, changed, or removed in Tana without your confirmation. To view proposals created from external tools like Claude, click **Browse** in the sidebar, then view **Chats**. Always check proposals before approving, especially updates to existing documents. See [Proposals](/learn/features/proposals) for how the review surface works. ## Tips - **Be specific about spaces**: if similar document names exist in different spaces, mention the space so Claude finds the right one. - **Reference meetings by date or title**: Claude can look up events by date range or by name. - **Ask for a specific kind of document**: a Bug or a Decision lands with the right fields, in the right space. ## Disconnect Tana To remove the connection in Claude Code: ``` claude mcp remove tana ``` In Claude Desktop or claude.ai, open **Settings**, then **Connectors**, and disconnect **Tana**. Disconnecting stops Claude from reading or writing to Tana. It does not delete any data already in Tana. You can also review or revoke access from your Tana account settings. ## Troubleshooting **Claude says it can't reach Tana.** The connection has likely expired or been revoked. Reconnect with the steps above. **Claude can't find a document.** Paste the document's Tana link straight into the conversation and Claude opens exactly that item, no searching needed. Otherwise, name the space, or reference the item by its type or exact title. Semantic search helps when you don't remember the exact wording. **A proposal didn't appear in Tana.** Proposals land in the organization you connected. If you have access to more than one, confirm you're looking at the right one. --- #### Walk into every meeting prepared URL: https://tana.inc/learn/guides/prepare-for-meetings Ask Tana to draft your meeting prep from your team's record, capture it as a reusable skill so it runs the same way every time, then hand that skill to an agent on a schedule so a prep is waiting before every meeting. {/* DRAFT for review. Teaching exemplar: the steps are the lesson. Uses meeting prep to teach the ask -> skill -> scheduled-agent escalation. Skills are the centerpiece; agent-building is handed off to agents-and-skills rather than duplicated. */} Good meetings start before the meeting. But prep usually means digging: who is this person again, what did we decide last time, what is still open, what should I cover. That is the scramble in the five minutes before the call, if it happens at all. This guide gets you out of that scramble, and along the way it teaches a pattern you will reach for constantly in Tana: do something once by asking, capture it as a [skill](/learn/features/skills) so it runs the same way every time, then hand that skill to an [agent](/learn/features/agents) so it happens without you. Meeting prep is the example. The climb from a one-off ask, to a reusable skill, to a scheduled agent is the real lesson. ## From one ask to automatic prep Open a [chat](/learn/features/chat) and say who and what: _"Prep me for my 2pm with Acme. Pull what we decided last time, any open issues they reported, and who I am meeting."_ Tana drafts the prep from your team's own record: the past meetings with these people, the decisions that came out of them, the issues still open, and the people involved. With your CRM [connected](/learn/guides/connect-hubspot) it folds in the deal and contact history. Because it works within your permissions and only from what your team actually recorded, the result is grounded in real context, not a generic briefing. Spend a moment getting the shape right, which items to pull and how to lay them out, because that shape is the thing you are about to make permanent. You do not want to retype that prompt before every meeting, and you want the prep to come out the same way each time. That is exactly what a [skill](/learn/features/skills) is: reusable AI instructions you write once and run whenever you need them. The fastest way to make one is to ask. Right after Tana gives you a prep you like, tell it to _"save that as a skill called Meeting prep"_, and it turns the instructions it just followed into a reusable skill, no setup and no prompt engineering. You can still open and edit it by hand, or build one from the create menu, but describing it to the AI is the normal path. Either way a skill is just a title, a short description, and the recipe in its body: pull the last meeting's decisions, the attendee's open issues, who they are, the CRM history if it is connected, laid out as talking points. From then on, anyone runs that prep by typing **/** in a chat and picking the skill, then naming the meeting to prep, instead of describing the whole thing from scratch. Here is why this step matters more than it looks. The prep becomes **consistent**, every meeting gets the same thorough treatment, with nothing dropped in a rush. And because skills are **shared**, a teammate covering your call runs the identical skill and walks in just as prepared. A skill is how you capture a good way of working once so the whole team has it every time, which is the same reason you would build one for filing a bug or drafting a recap. The last move is to stop running it at all. An [agent](/learn/features/agents) is an assistant you configure for a recurring job and then equip with skills, and "prep me for my meetings" is the textbook case. You describe the agent in plain language and turn on your meeting-prep skill in its Customize panel, the same skills-and-integrations setup every agent uses. Then give it a [schedule](/learn/features/agents#schedules): pick weekday mornings and a time in your timezone, and the agent runs while you sleep. Each morning it preps the day's meetings and leaves a prep doc for each one in your Library, so the work is waiting for you to review rather than assemble before the first call. Building and equipping that agent is its own short flow, walked through end to end in [Create AI agents with skills and integrations](/learn/guides/agents-and-skills). The point here is the shape of the escalation: the skill you wrote in the previous step is exactly what the agent runs on the schedule. ## The prep gets sharper over time One reason to invest the few minutes this takes: the prep compounds. In your first week with a new account it is mostly introductions. A month of meetings later it is the full relationship, what was promised, what slipped, what to raise next, because the skill draws on a record your team keeps adding to. You build the skill and the agent once, and they keep getting better on their own as the team works in Tana. --- ### Tutorials #### Turn a customer call into product feedback URL: https://tana.inc/learn/tutorials/customer-call-to-product-feedback Run a customer call in Tana and walk out with structured product feedback your team can act on: the pain points, a customer-journey map, and screenshots and annotations of exactly where they struggled. {/* DRAFT for review. Specialized meeting tutorial. Centered on the default product-feedback skill (validated wow: screenshots + annotations) and the Customer Journey artifact (built-in). Capture pitched above the in-flux UX. */} Customer calls are full of signal and almost impossible to capture by hand. You are listening, asking, and scribbling at once, and the texture is gone by the time you write it up. In Tana, the call does that part for you. Run a customer call in Tana and walk out with structured product feedback your team can act on: a summary, the specific pain points, and screenshots and annotations of exactly where they struggled. It uses one of Tana's default skills, so there is nothing to set up to get your first one. ## Run the call Start a meeting in Tana and join. The richest signal comes from the screen, so ask them to share their workflow, or share your product and watch them use it. Tana transcribes the conversation and reads the shared screen, so you can stop scribbling and actually listen. Run the call you were going to run. As they walk their process or react to a feature, Tana follows both what they say and what is on screen. It captures the shared screen, never anyone's camera. Run the default **product feedback** [skill](/learn/features/skills), from the Skills button in the [meeting](/learn/features/meetings) or with `/` in [chat](/learn/features/chat). Tana turns the conversation into structured feedback: a summary, the specific pain points, and screenshots with annotations marking exactly what they reacted to. This is the part that lands, the feedback is tied to the moment it happened, not reconstructed from memory. For the bigger picture, generate a customer-journey [artifact](/learn/features/artifacts): a before-and-after map of how they work today and where it could go. It is ready to share with the product team or a stakeholder as it is. Review what Tana produced, it arrives as a [proposal](/learn/features/proposals) you accept, then file the pain points as typed insights in your team's space, so the whole product team sees them, not just you. From there, push a specific item to [Linear](/learn/guides/connect-linear) when it is ready to act on. ## What you walked away with A customer call used to leave you with messy notes and a write-up to do. Now it leaves you with product feedback the team can act on, the pain points, the journey, the exact screenshots, shared minutes after the call. And it compounds. Because every call files into the same shared insights, the patterns surface on their own: the third customer this month to hit the same wall, the feature five interviews keep circling. Your team's understanding of its users gets sharper with every conversation, and the AI draws on all of it. --- #### Turn a design review into tracked work URL: https://tana.inc/learn/tutorials/design-review-tracked-work Run a design review in Tana and leave with your tracker already updated: feedback routed to the right existing issues, new changes filed, each tied to what was on screen and assigned to an owner. {/* DRAFT for review. Specialized activation tutorial (design review -> tracked work, primarily updates to existing issues/spec). Pitched above the in-flux meeting-capture UX: experience and loop level, not click-by-click. Payoff screenshot deferred until capture settles. Assumes a tracker (Linear/GitHub) or Tana tasks; pairs with the Connect GitHub and Linear guide. */} A design review produces a lot of feedback and almost no record of it. Half lands in scattered notes, half stays in someone's head, and afterward someone has to translate "we discussed some stuff" into actual issues, usually by hand, usually later. In this tutorial you run a design review in Tana and leave with the work already tracked. Most feedback lands on issues that **already exist**, the issue for that screen, the spec, the PRD, updated in place with the new comment and the screenshot. Genuinely new changes get filed as new issues. Either way, the backlog is current before the call ends. It works best with your tracker connected (Linear or GitHub), so the work lands where your team already lives. See [Connect GitHub](/learn/guides/connect-github) and [Connect Linear](/learn/guides/connect-linear). You need a design you can share on screen, a Figma file, a prototype, a build. ## Run the review Start a meeting in Tana, join, and share the design on your screen. Tana reads the screen you share and transcribes the conversation, so you do not narrate for the record or take notes. Walk the design and react out loud: "the empty state needs a call to action," "this spacing feels off on mobile," "the error copy is too harsh." Move between screens freely. Tana follows both the conversation and what is on screen, so each comment stays connected to what you were looking at. It captures the shared screen, never anyone's camera. As you go, or when you are done, have Tana pull the feedback together. It searches your existing issues and specs and, by default, updates the work each comment belongs to: a note added to the issue for that screen, a new acceptance criterion on the spec, the screenshot of what was on screen attached. Only genuinely new changes become new issues. You get updates to real work, not a pile of duplicates, and you do not have to set anything up for this to happen. If a review is squarely about one tracked doc, you can pin that doc to the meeting as a strong hint, but it is optional. Tana looks for the matching work on its own. Everything arrives as a [proposal](/learn/features/proposals), so nothing changes until you say so. For each item you see whether it updates an existing issue or creates a new one, where it lands, and who it is assigned to, and for updates you see the before and after. Tidy the wording, drop anything that was just thinking out loud, reassign as needed, and accept. By the end of the call, the work is where your team already looks. The designer and engineers open their tracker and the feedback is already on the right issues, assigned, with the visual context attached. Nothing to write up, nothing to re-file. {/* TODO screenshot (deferred until meeting-capture UX settles): the review's proposals, showing a mix of updates to existing issues and a new one, each with its screenshot. */} ## What you walked away with The review used to end with a vague sense of "lots of feedback" and someone left to turn it into issues later. Now it ends with the tracker already updated: existing work enriched, new work filed, each item tied to exactly what was on screen. The part a note-taker cannot do is the routing. Because Tana knows your issues and your spec, the feedback lands _on the right things_ instead of becoming a fresh list you have to reconcile against work that already exists. That is the difference between a meeting that produced talk and a meeting that moved the work forward. --- ## Blog ### Tana auto-detects external meetings URL: https://tana.inc/blog/tana-auto-detects-external-meetings Published: 2026-08-25 Tana notices when you are in a Zoom, Teams, Meet, or Slack call that never made it onto your calendar, and asks whether to capture it. One click turns it into a transcript, a summary, and outcomes. Plus presenting a real document instead of your screen, Tana in Slack, and our SOC 2® Type 2 report. The meetings that go uncaptured are rarely the ones on your calendar. They are the call someone pulls you into, the Slack huddle, the Zoom link in a DM. Tana now notices when you are in one and asks whether to capture it. One click turns it into a transcript, a summary, and outcomes you review in Tana. Presenting a document no longer means sharing your screen, and Tana answers questions in the Slack thread where they came up. A call you hold in Zoom, Microsoft Teams, or Google Meet takes one click to capture, and no longer starts with a setting. It records through your computer's own audio on the Mac and Windows desktop apps, so the list does not stop at those three. Any tool your machine plays sound through works the same way, and the call stays where it is. The meeting does not have to be on your calendar at all. Tana notices meeting-like activity from your microphone and speakers and asks, so the call someone pulled you into is one click from being captured rather than gone. It only ever offers. Nothing records until you press **Capture Meeting**. [What it looks for](/learn/features/meetings#detecting-a-meeting-you-did-not-schedule), and how to turn it off, are in the docs. For the ones that are on your calendar, a window still opens a couple of minutes beforehand, and **Open & Capture** starts the recording and opens the meeting together, so a blocked pop-up never leaves you sitting in an unrecorded call. Pause and stop live in that window, and for a call that appears out of nowhere you can start a capture from Today. Pick a document or canvas from the **Present** palette in the meeting toolbar and it takes the center of the meeting for everyone. It is not a picture of a document. It is the document, rendered on each person's own screen. That difference is the whole point. Text stays sharp and selectable at any window size instead of arriving as a compressed video of your monitor. People can scroll ahead to read something again and press **Follow** to come back to you. Nobody has to ask whether they can see your screen, and you are not hunting for the right window while everyone waits. You edit it live while the room watches it change, so the notes you take during the meeting are simply the meeting's notes, in Tana and attached to the event the moment the call ends. Guests read the same live page as everyone else. What was on screen survives the call too. The digest records the document you presented, so reading back through a meeting shows which one was on the table at which point. In a Tana call, screenshots from a shared screen sit next to the discussion they belong to. See [how presenting works](/learn/features/meetings#presenting). Meetings produce work that points somewhere else, and following one of those threads used to mean losing your place. Right-click most things in Tana now, a meeting tile, a pinned document, an accepted outcome, and open it in a new tab. The desktop app adds two more places to put it, the side panel beside what you are already reading or a window of its own. Modifier keys do the same without the menu, and on desktop a number key jumps straight to a tab. Search opens as a centered dialog now, wide enough to read results as a table. Not every question is worth opening another tab for. Mention **@tana** in a Slack thread and it answers in place, reading the thread and the files shared in it, so you can drop in a screenshot and ask about it without moving anything. Type **/tana** in a channel or a DM and you get a Tana meeting link back in the same place, so a thread that has outgrown typing becomes a call. It goes the other way too. Give an agent a morning schedule and its work arrives in Slack looking like something a person wrote. Links, bold, bullets, and headings survive the trip, and a link to a Tana document arrives as something your teammates can click. The two directions are set up separately. An admin installs the Tana app once for the organization, which is what **@tana** and **/tana** run on, and you connect your own Slack account in Settings, Integrations to let your agents post as you. [The Slack docs](/learn/features/integrations#slack) have both. A type's fields now show up as columns on its entries table and its workflow board, and as a row of filter pills beside the search box. Narrow proposals to one customer, tasks to one owner, or anything dated to a range, by picking values rather than writing a query. The filters live in the URL, so a filtered board is a link you can send someone, and opening an entry and coming back leaves the filtering exactly as you left it. The AI reads the same fields. Ask for "open proposals from customer Nordfish" or "feature requests in Product area Billing" and it resolves the constraint in one scoped search instead of listing every candidate and narrowing them one at a time. ## More in this update The transcript lives in one place, grouped by speaker under time markers that stay put as you scroll. Click a line or shift-click a range to copy it, or copy the whole thing with nothing selected. Opening it during a call keeps you in the meeting. Reviewing what the AI proposed and correcting it are the same screen. Edit a value, add a field, or remove one while the old version stays struck through beside the new. The meeting wrap-up leaves a field blank rather than guessing at it, and quotes only what was said in the meeting it is summarizing. A record that looks complete gets trusted, so it has to be. Meeting links are now `meet.tana.inc` plus a short code. Short enough to read out on a call, and invites carry the same one the meeting card shows. Anyone can put an agent on a schedule, including agents shared with them. Schedules stay private to whoever made them, and the opening message is optional. Ask the AI to find one of your own agents by what it does, like "find the agent that does customer research". Attach a skill to a message with a slash and it runs straight away, using whatever you typed as its input. Admins can change a member's role between Member and Admin after inviting them, without starting over. Screen shares get their own featured region during a call, and clicking one enlarges it. Raised hands come to the front of the tiles. A stalled calendar now says so, with a banner above Today and a button that takes you where you can reconnect. Large PDFs attach to a chat without failing, so a long report reads like any smaller file. With GitHub connected, the AI can list your organization's members and their roles, and who has access to a repository, so you can ask who owns something before you assign it. **Send to Linear** and **Send to Jira** create the issue now, instead of reporting themselves unavailable. Device settings name the microphone the call is really using, and plugging in a headset mid-call moves the call onto it. Synced events keep their end times, and all-day events pick up later changes, even when Tana cannot identify the organizer. Copying a link while a document is open in the preview overlay copies that document, not the page behind it. Admins can upload an organization logo under Settings. It replaces the generated initials everywhere the organization appears. AI chat spends fewer tokens on the same work, so you are less likely to meet a usage limit. The full list of what shipped lives in the [changelog](/changelog). ## SOC 2® Type 2 Your security team's favorite question has a one-word answer now, and the word is yes. Prescient Assurance LLC examined the controls behind Tana, and the SOC 2 Type 2 report is ready. Request it on our trust center at [trust.tana.inc](https://trust.tana.inc), along with our security practices and the current sub-processor list. Tana Outliner is not in scope. DPA requests go to privacy@tana.inc. ## Try this Capture picks up the meetings you did not plan for, which means the loose ones now leave a trace too. Ask Tana what came out of them. --- ### Bring every meeting into Tana URL: https://tana.inc/blog/bring-every-meeting-into-tana Published: 2026-07-08 Capture meetings from any tool, keep your calendar in sync both ways, and pin your own sidebar. Plus a sharper meeting view and a calmer voice agent. This update is about Tana fitting the way you already work. Your calendar now syncs both ways, capturing the meetings you hold in other tools is smoother and more reliable, and the sidebar is yours to arrange. The meeting itself, and the voice agent in it, got sharper too. Events you create and manage in Tana now sync out to your connected calendar. Make an event in Tana and it appears in Google Calendar or Outlook moments later, and changes to its title or time, and deleting it, propagate back too. New connections write Tana events to your primary calendar automatically. If you connected a calendar earlier, turn it on by choosing which calendar to write to in [calendar settings](/learn/features/meetings#calendar-sync). Capturing a meeting held somewhere else now works better. Because it records the audio your computer plays, it works across any meeting platform rather than a fixed set, and it is lighter on your machine and more reliable. A small always-on-top window opens a couple of minutes before a scheduled meeting so you can start recording in one click, with pause and stop controls right there. You can also start a capture from the Today page for a meeting that is not on your calendar. If a teammate is already capturing the same meeting, everyone else sees it so it is not recorded twice. It is rolling out as an opt-in. On the latest desktop app, turn on **Use experimental meeting capture** in Settings, Preferences. See [capturing external meetings](/learn/features/meetings#capture-with-system-audio) for the details. The sidebar is now yours to shape. Instead of listing every space and agent automatically, it keeps a clean set of primary links and a **Pinned** area you fill with the things you actually reach for. Pin any document, space, or agent, drag to reorder, and right-click to create a new section and group your pins. Your full set of spaces is still there in the Library whenever you need it. The meeting view is now one calm column that carries the work instead of a stack of documents. What the AI captures during the call is sorted into two clear places: items wait under **For Review** until you accept them, with keep and reject buttons right on the card, and accepted ones move to **Outcomes**. During the call the live digest steps back to a slim status line, so the shared focus takes center stage, and the full timeline stays one click away. When the meeting ends you get an editable, shareable summary with a Summary and Timeline toggle. See [how outcomes work](/learn/features/meetings#review-and-use) for the details. ## More in this update The voice agent has a calmer, more ambient look and clearer audio, and you can set a default voice for Tana's built-in agents in Organization Settings. Invite anyone to a meeting straight from Tana. Add teammates by name or type any email address, including people outside your organization, and everyone you add gets a real calendar invite in their inbox. Transcription follows your organization's language across meetings, external capture, the voice cursor, and dictation, while still picking up other languages spoken in the room. Guests in a Tana meeting can ask the AI questions in their own private chat, separate from the shared meeting chat. It answers but cannot create or change anything. Say "Hey Tana" and give a command in the same breath, like "Hey Tana, file that as a bug", and the agent acts on it instead of just greeting you. Tana opens faster, especially on slower connections and right after an update. Incoming-call notifications now reach you in the browser, not just the desktop app, and take you straight to the meeting. The full list of what shipped lives in the [changelog](/changelog). ## Try this --- ### From conversation to action URL: https://tana.inc/blog/from-conversation-to-action Published: 2026-06-19 The AI can turn a conversation into a calendar event, guests can join the shared meeting chat, and a new Capture as action grabs the exact part that mattered. Plus Meet now or later from Today, diagrams inline in chat, and more. This update is about the conversation, and everything you want to do with it. The AI can turn it into a calendar event, guests can join the shared chat, and you can grab the exact part of a meeting that mattered without keeping the rest. Ask in [chat](/learn/features/chat) and the AI can now [propose](/learn/features/proposals) a calendar event. Give it a title and a start time and it drafts the rest: an end time, time zone, location, description, and the people to invite. Adding participants keeps the event to that list. You stay in control. It is a proposal you review before anything lands, and the AI still will not change the time, title, attendees, or meeting link of an event that already exists. Meetings have a shared chat, Chat with everyone, and now guests are in it too. They can send and read messages right alongside your team, the simplest way to bring an outside collaborator into the conversation as it happens. To keep your organization's data private, the AI steps out of the shared chat whenever a guest is present, with a short note explaining the pause. Your own private meeting agent keeps working as usual. A meeting can cover a dozen different things. Capture as lets you act on just one of them: hover any row in the live [digest](/learn/features/meetings#transcription) and turn that section into a note, a task, a typed item, or an [artifact](/learn/features/artifacts) like a screen-share recap. Scoped to that one part, the AI works from focused context, so what comes out is about that decision or action, not the whole conversation. Need the exact words instead? The same hover menu copies that segment of the transcript, grouped by speaker. The Meet button in [Today](/learn/features/home) splits in two. Meet now starts a meeting straight away and opens a "Who are you meeting with?" popover, where you can copy a guest link, add colleagues in one click, share your screen solo, or start a voice chat with an agent. Meet later opens the scheduling popover instead, for picking a time. One button, two clear intents: start talking now, or put it on the calendar for later. ## More in this update **Diagrams render in [chat](/learn/features/chat) replies.** When the AI includes a diagram, it now renders inline once it is complete, instead of flashing raw markup as it streams. **Open a [task](/learn/features/tasks#the-tasks-view) in a peek.** Click a task's title in the Tasks view and it opens in a popover over the list, so you can read or edit it without leaving your tasks. **A date chip on an event's location pill.** It reads Today, Tomorrow, Yesterday, or a short date, and clicking it jumps to that day on the calendar. **Know when a [type](/learn/features/types#archiving-a-type) is archived or restored.** A notification tells you when someone in your org archives or restores a type, with a link to its panel. **Filter any [Library](/learn/features/interface#more) section.** Created by and Visibility pills now sit on every section page, saved in the URL so the view is bookmarkable. **Your own [skills](/learn/features/skills), front and center.** The Skills section and search now list the skills you and your team have made, with Tana's built-in ones filtered out. **Open org links in the [desktop app](/learn/features/apps).** Click a link to your org's Tana in a browser and it asks whether to open in Tana or stay in the browser, then remembers your choice. Set it from Preferences in the web app. **Find events you organized in [Created by](/learn/features/interface#filters).** Calendar events are now attributed to their organizer, so filtering by Created by surfaces the meetings you set up. The throughline this time: less time spent capturing the meeting, more spent acting on it. The [changelog](/changelog) has the rest of this release, including a handful of call and sign-in fixes. ## Try this The AI can schedule a meeting and prep your team for it in one go, drawing on what it already knows about your work. Drop this into a chat and it drafts the event and a prep doc for you to approve: --- ### Bring your own tools URL: https://tana.inc/blog/bring-your-own-tools Published: 2026-06-11 Tana reaches further this update: its AI can use the tools on your own MCP servers, the Tasks view groups and filters your work into bookmarkable slices, and a new search popover gets you to anything in Tana from wherever you are. This update is about reach. Tana's AI can now use the tools on your own servers, a reworked Tasks view groups and filters your work into slices you can bookmark and share, and a new search popover gets you to anything in Tana from wherever you happen to be. Tana already connects to GitHub, Slack, Linear, Jira, and HubSpot. Now you can connect anything else that speaks the Model Context Protocol, your own internal server or a third party's, and its tools become available to Tana's AI. Add a server from settings and authorize it once over OAuth. Each connection becomes a [skill](/learn/features/skills), so its tools only enter a chat when that skill is in scope, and you can allow or deny each tool from the skill's panel. Your connected servers stay private to you. It runs both ways, too. Tana is itself [an MCP server](/learn/features/mcp), so Claude, Cursor, or any MCP client can work with Tana from the other side. See [Custom MCP servers](/learn/features/integrations#custom-mcp-servers) for the full picture. The [Tasks](/learn/features/tasks) view is now built around filter pills, assignee, creator, type, and space, that live in the URL, so any slice of your tasks, everything assigned to you, what you have delegated, only your open issues, becomes a link you can bookmark or share, and the browser's Back and Forward walk your filter history. Group the list by type, status, or assignee, with a "+" on every group heading that adds a task already matching that group and your active filters. With nothing filtered, the view shows your most recent tasks rather than an empty screen. [Search](/learn/features/interface#search) now opens as a popover right where you are, including while you are writing in a document. Start typing for full-text search across Tana, or open it empty to browse. When nothing is typed, the popover shows the [Library](/learn/features/interface#more) as a browsable index: each kind of thing, meetings, tasks, types, chats, and the rest, in its own section with a "See all" page. It is the fastest way to land on something without losing your place. Reach for it with Cmd+P on macOS, or Ctrl+P on Windows and Linux. ## More in this update **Filter by who can see it.** A new visibility filter in [search and the Library](/learn/features/interface#search) narrows any list to what is private to you, shared with you, restricted, open, or shared by link. **Set an [outcome's](/learn/features/meetings#outcomes) state from the meeting.** Outcome tiles show a state control inline for anything with a workflow, a plain task or a typed item like a Bug, so you can move it along without opening it. **See where any item lives.** The location button on a panel now names the [Library section](/learn/features/interface#more) an item belongs to, like "Meetings" or "Docs", and links straight there. **Recent meetings, front and center.** The [Meetings preview](/learn/features/interface#more) on your Library and space home leads with recent meetings, with a dropdown to switch between recent, upcoming, today, and past. **[Reopen an archived type](/learn/features/types#archiving-a-type).** A "Recently archived" list in settings keeps every type you have archived, so you can find one and unarchive it. **A [title chip](/learn/features/interface#panel-title-chip) when you scroll.** Scroll down any panel and a compact chip with the item's icon and type appears in the toolbar. Click it to jump back to the top. **[Dictation](/learn/features/chat#voice-input) keeps your last words.** Send a dictated message the moment you stop speaking and the tail of what you said now makes it in, instead of getting dropped. **Copy a long prompt for a [coding tool](/learn/features/integrations#coding-tools).** When a hand-off prompt is too long for a tool's launch link, Tana offers a "Copy full prompt" button so you can paste the whole thing. **Open a [video](/learn/features/docs#images-and-attachments) in its own player.** Opening a video item now plays it in a dedicated player panel with the usual toolbar, instead of a generic view. **[Transcription](/learn/features/meetings#transcription) spells your team's names right.** Meetings and dictation now lean on your organization's name and your teammates' names, so "Tana" and the people in the room come through correctly. That is the gist. The full list, every new, improved, and fixed item, lives in the [changelog](/changelog). ## Try this The new search popover and Library are about getting to anything fast. Let the AI do the gathering: drop this into a chat and it pulls a scattered topic into one place. --- ### Tana takes it from here URL: https://tana.inc/blog/tana-takes-it-from-here Published: 2026-06-05 Tana does more on its own: meetings wrap themselves into outcomes, there's one place to create without leaving the call, and your tasks and mentions show more at a glance. Plus a unified Library, Mermaid diagrams, and Outlook calendar sync. This update shares a thread: Tana does more on its own. Meetings wrap themselves into outcomes the moment everyone leaves, there's one place to create or capture without leaving the call, and your tasks and mentions show more at a glance. Once everyone leaves a call, Tana runs the [wrap-up](/learn/features/meetings#outcomes) on its own: it proposes outcomes and writes a summary, with no manual step. Want them while you are still in the call? Hit Wrap up. When you pin a document to a meeting, the wrap-up treats it as the strongest signal of what the meeting was about, and updates that document rather than creating a new one. The meeting toolbar is simpler. Capture, Skills, and Create collapse into a single create menu, with a spyglass beside it to find and pin things. From that one menu you can turn the last minute, or the whole meeting, into a typed item, make a private note or task just for you, or run a skill or summary. One place, less hunting. Browse is now Library, the home for everything that is not pinned to a space, and it shares one filter bar with [Search](/learn/features/interface). Filter pills jump you to meetings or chats, or narrow by who created an item, and your spaces and types each get a dedicated section so they are easy to reach. Once you have found something, the new ["Send to"](/learn/features/docs#send-to) menu routes it anywhere: start a chat about it, run a skill, send it to an integration, pin it to a meeting, copy its link, or export it as Markdown. A task now shows a state icon wherever it appears, filling like a pie as it moves through its workflow, so you can tell where something stands without opening it. Give any [type](/learn/features/types#workflows) a workflow to make it task-like. Inline mentions carry more too: a task's state, a person's avatar, an image thumbnail, or the item's type. Hover one for a preview, or right-click to pin it to a date, assign it, or change its status. You can now add diagrams to any document. Type "/mermaid", ask the AI in chat, or set a code block's language to Mermaid, then switch between a live preview and the source. When the AI writes a diagram, it self-corrects unsupported syntax on its own. ## More in this update **Connect your Outlook calendar.** Microsoft and Outlook calendars now sync into Tana, so those meetings show up alongside the rest. **Activity is now a dashboard.** Promoted into the sidebar, it shows your notifications and recently updated items side by side, with an unread count. **A guided welcome session for new team members.** New workspaces open with a live, guided meeting: an agent tuned to your work sets up spaces and types with filled-in examples, and walks you through the essentials. **[Agents](/learn/features/agents) know what time it is.** Agents are now timezone-aware, so "today" matches your own calendar day and a [scheduled routine](/learn/features/agents#schedules) set to 09:00 in Oslo lands on the right Oslo morning. **Reassign right from the board.** Kanban cards carry an assignee avatar. Click it for a searchable people picker and reassign without opening the card. **Paste a link, get a mention.** Pasting a link to a Tana item now drops in an inline [mention](/learn/features/docs#mentions), not a raw URL. **Meeting summaries open with an illustration.** A wide pen-and-watercolour image, drawn from what the meeting was about, now tops each summary. **Muted by default in larger meetings.** Join a call that already has five or more people and you start muted, so you don't cut in. **[Archive a type](/learn/features/types#archiving-a-type).** Archive a type and it drops out of every menu, picker, and search right away, while existing items keep it with a muted "(archived)" label. **[Restore a deleted document](/learn/features/docs#deleting-a-doc).** A soft-deleted item now shows a "Restore document" button, so you can bring back something removed by mistake. That is the gist. The full list, every new, improved, and fixed item, lives in the [changelog](/changelog). ## Try this Diagrams are the easiest new feature to try yourself. Drop this prompt into your chat and see what Tana draws: ---