Tana now offers a fully integrated AI meeting notetaker as part of the desktop experience. You can use it to transcribe meetings, generate summaries, and link notes directly to your work - without needing to add a bot in the call.
Tana now offers a fully integrated AI meeting notetaker as part of the desktop experience. You can use it to transcribe meetings, generate summaries, and link notes directly to your work - without needing to install a plugin or add a bot.
The meeting notetaker works for both calls and in-person meetings by recording system audio, and can also be used for lectures, interviews, and other audio sources. The feature is available on both Mac and Windows through the Tana desktop app and supports over 60 languages with automatic language detection.
Click the microphone button on any note tagged with an audio-enabled supertag.
Click the desktop notification if Tana detects a meeting or you've synced your Google Calendar to Tana.
Use the command line with Cmd+K (or Ctrl+K), choose “Start meeting with...”, and select the supertag you want to use (all audio-enabled tags will show).
When meeting transcription starts you’ll see a floating recording menu, and a recording indicator in the sidebar. This gives you quick controls to pause for privacy, or end the meeting and start post-processing. Clicking the recording indicator will open the meeting note in Tana.
You'll see a Transcribing... button on the meeting that you can click to open and see the live transcription in source material. This button will also be available when you pause or stop the transcription so you can revisit the transcription at any time.
Tana AI chat is available on all notes. To chat with a meeting note: - Press space on an empty line inside the meeting - Or click ✨ Ask AI about this content in the top right
If real-time transcription stops due to network issues or vendor outages, you can retranscribe from the captured audio file using the Retranscribe from recordings command available in context menu on right-click or through the command line.
Click on the user icon > Tana Labs > Transcript backup audio recording. All your transcriptions will now automatically capture back audio, so you can re-transcribe if you experience problems with network or language output.
Audio recordings are automatically captured and stored to enable re-transcription when needed:
Audio files are stored securely in your Tana cloud storage, not locally on your device
Recordings have a technical limit of 59 minutes. Longer sessions will create multiple audio files (e.g., a 3-hour meeting will generate 3 separate recordings)
Starting and stopping transcription multiple times creates multiple audio recordings
If network connection is permanently lost during recording, audio files will not be persisted after 5 retry attempts
Audio files are not directly accessible through the interface
Tana supports realtime transcription in over 60 languages. By default, transcription uses auto-detect, which can dynamically switch between languages during a meeting.
To change this:
Shift-click the microphone button before starting the transcription,
Click the globe icon in the transcription view during, or
Set a default language using the Set default transcription language to... command
The summary will be generated in the same language as the spoken language. You can translate the summary by using AI chat.
You can add context to improve the quality of transcription for difficult words or names. This is configured in the Realtime transcribe audio command (automatically added when you enable audio on a tag):
Add the Transcription context parameter to the Realtime transcribe audio command under "Compact menu".
Include names or other vocabulary manually, or reference a field (e.g. Attendees) to include dynamic context from the meeting.
NewRe-transcribe from audio recording. If a transcript stops because of network drops, or you want a different language as output, you can now re-process it from backup audio. This is optional, and can be enabled under Tana Labs in user settings as "Transcript backup audio recording". ()
ImprovedBetter transcription model which supports 60 most common languages, very good at recognizing language switching and automatically choosing correct language. Used both for live transcription in app and from mobile capture. Non-supported languages in app fall back to Whisper (if you set it with Set default transcription language command). ()
FixedFixed transcription formatting and improved accuracy for Chinese language input. ()
Yes, Tana now offers a fully integrated bot-less AI meeting notetaker as part of the desktop experience. You can use it to transcribe meetings, generate summaries, and link notes directly to your work - without needing to install a plugin or add a bot.
How do I change the tags used in the meeting agent and calendar integration?
Dec 02, 2024
Supertags are created or assigned as part of the setup for your meeting agent/calendar integration. Here's a quick guide on how to change these tags post-setup.
Supertags that are used during calendar sync are listed under Classification of events in the Google Calendar settings:
Go to Account settings > Settings > Google Calendar settings > Classification of events
For more on what the different event types mean, see here.
When you change a tag, it needs to use the same fields that are set up for the calendar integration, otherwise the information will get lost. To check which fields are used during sync, go to Google Calendar settings > Advanced > Raw configuration:
The raw configuration defines all the fields that Tana will sync specific calendar event information to.
Supertags are used when processing the transcript to create meeting notes. These supertags are set during setup and can be changed by going into the command node that controls the meeting agent and the transcription process.
First, you must navigate to the text processing agent command for your meetings. For instructions on how to find this, go here and scroll up: Meeting agent > Text processing agent command
You should now have the Text processing agent config open. Here you will see parameters like "Tags used for X", which is the list of supertags used when processing meetings.
To change the tags, you simply add or replace tags from these lists.
However, there are several things to consider when you add/replace from these lists:
Study existing tag: Open the config for the tag you want to replace, and study how it's set up. Try and identify dependencies like specific fields that are also used in the meeting config. Try and make sure your replacement tag replaces as much of the original one as possible.
Set Base type: Make sure the new supertag you add has the appropriate Base type defined, and that it's suitable for the type of object it is:
Items typically use base tags: Project, Topic, Event
Entities typically use base tags: Person, Organization, Location
Action items typically use base tag Task
Decommission the old tag: Make sure the tag you're no longer using is merged with the one you're using, or deleted, to prevent clutter and confusion
There are likely more things to consider, let us know if we missed any via Rate this article 👍👎.
Does the GPT log monitor show AI credits usage by the Meeting agent?
Sep 12, 2024
Only partially; the transcription of the meeting does not use OpenAI and is therefore not logged in the GPT log monitor.
Meetings use a service called Soniox and we are working with them to expose the AI credit cost of individual meetings in a simple way, but currently it's not possible.
The post-processing of the meeting transcript is however done using OpenAI and will be visible in the GPT log monitor.
How can I revoke a meeting agent from a meeting I'm not hosting?
Sep 05, 2024
We are looking into creating a button that will make the agent leave a meeting if you press it. Until we have this in place, it is possible to force the agent to stop transcribing and kickstart the meeting notes processing if necessary.
To do this:
Run the command Debug node on the meeting node
Set the Meeting bot status to Done
The agent should now have stopped transcribing, and the text processing agent should start running.