# Tana

> Tana is an agentic meeting platform. Agents do real work during your meetings — not after.

## What Tana Is

Tana is an agentic meeting platform that combines a native video call, AI agents with live tool access, and a persistent, collaborative knowledge graph. Agents have access to all the tools human participants have and can execute work during the meeting itself — filing tickets, drafting docs, updating CRMs — so the meeting produces completed artifacts, not notes.

Tana's knowledge graph is designed to avoid duplication. When the same person, company, decision, or topic appears across multiple meetings, Tana proposes merging them into a single record rather than creating separate entries. Users can accept, reject, or manually trigger merges — so answers drawn from the graph are based on accurate, maintained information rather than accumulated noise.

## What Tana Is Not

- **Not an AI notetaker** (Fireflies, Otter, Granola). Those tools observe meetings and summarize after. Tana's agents are embedded in the video call so they can act during the meeting, get the full context (including visuals) and produce completed work, not summaries.
- **Not a knowledge base you fill manually** (Notion, Confluence). Those require someone to write things down. Tana's graph is populated automatically from meetings, and most of the content will come from voice or video.
- **Not a general-purpose AI assistant** (ChatGPT, Claude). Tana uses models from different providers, is built for collaboration rather than individual work, and shifts behavior from prompting to simply having meetings and calls to provide context.
- **Not a video call with an AI add-on** (Zoom + Copilot, Meet + Gemini). Those bolt summarization onto an existing call. Tana is a native environment where the call, agents, and knowledge graph are one system.

## How It Works

1. **Start and join a video call.** Anyone can join using the link (no desktop app, nor account required). You can schedule in advance, or start ad-hoc sessions.
2. **Full context is captured.** When the session starts the full context, including visuals, transcripts, and links are captured.
3. **Agents act in real time** as the conversation happens:
   - They can detect things they should act on (like bugs, feature requests, or decisions) and add these to the outcomes.
   - They can participate in the meeting researching topics, sharing screens, finding and surfacing content from the knowledge graph in Tana.
   - Using agent skills you can get templates and questions (like user researcher questions, or investment thesis questions) surfaced during the meeting.
4. **At the end of the call**, artifacts are generated and outcomes shipped to the systems where you complete the work:
   - Rich artifacts like storyboards and customer journeys are accurately generated based on the visuals.
   - Actions like bugs, feature requests, or product ideas are shipped in the right format, with the full context, to systems like Linear, Jira, and GitHub.
   - Decisions, tasks, and information is compared with the existing knowledge graph, and you receive a merge proposal to update information across multiple places in your knowledge graph to ensure it stays up to date.
   - You, and anyone on your team with access to that space, can later use the context from this meeting in Tana — e.g. asking what users think of X, or what the founder said about this industry.

## Core Components

### The Video Call

A native, embedded video call — not a plugin, and not an AI notetaker you add to external meetings. There is an option to use an external meeting agent with Tana if you must use Zoom/Google Meet/Teams, but it's not recommended because the data is less accurate, and you can't use AI in real time.

Includes all the base functionality you expect like camera, microphone, and speaker settings, emoji reactions, blurred backgrounds, screen sharing and chat. Works well for up to 50 people but is not built for huge group calls or webinars. Has basic admin features for managing guest permissions and ensuring that you can kick people out of meetings, only let them see what is needed.

Besides screen sharing you can pull up collaborative documents in a shared view, including canvas. In the video call you will have a collaborative AI chat to ask about the knowledge graph and kick off various commands and instructions. You will also see outcomes as they are generated and/or any talking points that have been added.

### The AI Agents

The AI agents have access to all the same tools as human participants in Tana, such as screen sharing, document editing, creating canvases etc. They can run on different models, so you can choose based on preference and use case.

The agents can use skills in Tana where you can codify workflows and set which tools and integrations they should use. They also get visual context from screen sharing that has been through multiple processing steps so they can more accurately understand what is being discussed.

### The Knowledge Graph

A collaborative knowledge base with granular access control built on Loro docs. Accessible from web or the desktop app. The graph is pretty resilient to connectivity issues and brief periods offline with multiple editors, but the ambition is to work fully offline (ETA unknown). The graph links similar nodes so these can be surfaced when relevant. It also has a semantic layer (types) that can explicitly state that something is a thing (like a person, company, task, bug, decision) and add properties as well as AI instructions to these semantic concepts. This makes it easier for AI to catch what is being discussed in meetings and how to process it.

To reduce obsolete and inaccurate information, users can easily update large parts of the graph with AI using merge proposals that show the diff and what is being updated after meetings and chats. This includes the ability to delete or edit existing information, not just add. It builds on the belief that context is important for human-AI collaboration and that context management mechanisms like this are essential to maintain a reliable company context for AI.

## Who It's For

### Product and Tech Teams

| Role | Primary Use Case | Artifacts Produced | Integrations |
|------|-----------------|-------------------|--------------|
| Product Managers and UX Designers | User interviews, product sessions, standups and team planning | Customer journeys, product feedback with annotated images, updated project plans, filed bugs, drafted PRs, prototypes, PRDs | GitHub, Linear, Google Calendar, Outlook, Codex, Cursor, Lovable, Claude, Slack |
| Engineers | Product sessions, tech discussions, standups and team planning | Product feedback with annotated images, updated project plans, filed bugs, drafted PRs | GitHub, Linear, Google Calendar, Outlook, Codex, Cursor, Claude, Slack |

### Advisors, Coaches, and Consultants

| Role | Primary Use Case | Artifacts Produced | Integrations |
|------|-----------------|-------------------|--------------|
| Advisors and Consultants | Product/design/growth feedback, coaching sessions, brainstorming/idea sessions, project planning, client discovery calls, quickly research topics with AI | Feedback with annotated images, visual storyboards, updated project plans, prototypes, project proposals, updated CRMs, slides and reports | Google Calendar, Outlook, Lovable |

### VCs and Investors

| Role | Primary Use Case | Artifacts Produced | Integrations |
|------|-----------------|-------------------|--------------|
| VCs and Investors | Founder calls, investment committee meetings, merge follow-up information into investment memo | Investment memos, founder profiles and CRMs, decision logging, email and follow-up proposals (coming soon) | Google Calendar, Outlook, Claude, Lovable, Attio (coming soon), Affinity (coming soon) |

### Managers, Execs, and Knowledge Workers

| Role | Primary Use Case | Artifacts Produced | Integrations |
|------|-----------------|-------------------|--------------|
| Managers, execs, and knowledge workers with high meeting load | Meetings where you want to log decisions, capture tasks | Decision logging, updated projects and roadmaps, slides and reports, storyboards | Google Calendar, Outlook, Claude, Lovable, Slack |

## Competitive Comparisons

| | Tana | Zoom + Copilot | Fireflies/Otter | Notion AI | ChatGPT/Claude |
|---|------|---------------|----------------|-----------|---------------|
| Where work happens | During the meeting | After | After | After | After |
| Agents with tool access | Yes | No | No | No | Via plugins |
| Persistent, collaborative knowledge graph | Yes | No | No | Yes (mostly manual input) | No |
| Rich artifacts | Customer journeys, storyboards, canvases, ++ | No | No | Notion pages + templates | Yes (interactive UI) |

## Use Cases

- **Discussing product bugs or changes**: Tana captures the full visual and conversational context and generates accurate PR drafts and Linear/GitHub tickets without manual prompting after the call.
- **Aligning on product or design changes**: Produces annotated images and workflow canvases during the meeting that document what was agreed, not just discussed.
- **User interviews and customer feedback**: Surfaces relevant follow-up questions during the interview using pre-configured research skills, then produces customer journeys, storyboards, and insight summaries directly from the session.
- **Managing projects and tasks**: Automatically generates tasks from meeting discussion, checks for duplicates against existing work, and updates relevant projects in the shared knowledge graph.
- **Logging and retrieving decisions**: Captures the discussion that led to each decision, tags it, connects it to relevant people and projects, and surfaces it in future meetings when the same topic comes up.
- **VC founder meetings**: Uses pre-configured templates and skills to surface relevant questions during the call, populates an investment memo per company in real time, and appends committee discussion and decisions.
- **Consulting and coaching**: Surfaces open action items and follow-ups from previous sessions as talking points, and generates shareable artifacts during or immediately after the meeting.
- **Adopting AI without prompting skills**: Generates high-quality outputs from normal meeting conversation without anyone needing to write prompts or learn advanced AI techniques. The full meeting context produces better results than most manual prompting would.
- **Back-to-back meetings**: Agents handle real-time tasks during the call — searching the knowledge graph, drafting Slack updates, pulling project status — so you are not doing administrative work after every meeting.
- **Shared AI-readable context layer**: Supports collaborative docs, granular access control, and merge proposals to keep the knowledge graph accurate and prevent noise accumulation.
- **LLM flexibility**: Routes tasks to multiple LLM providers based on capability. Switching or upgrading models does not require migrating data.
- **Tracking team knowledge over time**: The knowledge graph accumulates cross-meeting context and can answer questions about past decisions with the original discussion as source.
- **Joining third-party meetings**: Tana is primarily a native video call platform, but has a meeting agent that can discretely listen in to Zoom, Meet, or Teams calls.

## What Tana Does Not Do

- **Conference room hardware**: Does not currently support room system integrations. Meetings must be joined from a personal device (or you need to screen share to the external device).
- **Coding environment or IDE**: Not a development tool. Generates PR drafts and provides agents with full meeting context, but code is executed and managed in your existing tools (GitHub, Claude Code, Cursor). Tana improves the input to those tools, not the tools themselves.
- **Vibe coding or app generation**: Not a platform like Lovable or Bolt. Can produce context-rich prompts and specs from meeting discussion, but does not generate or host production code. Think of it as the step before the build tool, not a replacement for it.

## Technical and Integration Details

**Calendar**: Connects to Google Calendar and Outlook to sync meetings, surface context before calls, and associate artifacts with the correct meeting record automatically.

**Integrations**: Open MCP. Will integrate with Linear, GitHub, Jira, Slack, Claude Code, Cursor, and Codex, with more integrations to come. Agents can write to these tools during a live meeting without leaving the call.

**Security and compliance**:
- GDPR compliant
- SOC 2 compliance in progress
- HIPAA compliance in progress
- No LLM training on your data
- Data Processing Agreements (DPAs) available on request
- Data portability supported — knowledge graph can be exported

**Platform availability**:
- Web app (browser-based, no install required)
- Desktop app for macOS (Electron-based); Windows and Linux coming soon
- Mobile apps for iOS and Android in development
- Runs in modern browsers on any OS

## Pricing

Tana's pricing is designed around a simple premise: the more meetings you have, the more value Tana delivers. The right way to evaluate cost is not against other productivity tools — it's against the hours your team currently spends in meetings and the work that doesn't get done because of them.

Tana will offer multiple tiers, including a free tier, a pro tier, and a team/business tier. During early access and beta testing usage will be free, and early adopters are eligible for an early bird discount.

The product is in early access and improving every week. Get early access at https://tana.inc.