Skip to main content
A bot is a short-lived worker that joins one meeting and streams audio for transcription (and optionally persists a recording).

​
What You Get

Depending on your flags:
  • transcribe_enabled=true → transcript segments[] (live via WebSocket and/or post-meeting via REST)
  • recording_enabled=true → post-meeting recordings[] with playable/downloadable media
Recording persistence is separate from capture (Vexa must capture audio to transcribe). See timing + alignment details in:

​
Supported Meeting Platforms

The API is platform-agnostic. The only differences are the meeting identifiers you pass.
Platformplatformnative_meeting_idpasscodeNotes
Google Meetgoogle_meetabc-defg-hijNoOften requires host admission (waiting room)
Microsoft Teamsteamsnumeric ID (e.g. 1234567890123)Yes (?p= value)passcode is required for most Teams links
Zoomzoomnumeric ID (e.g. 12345678901)Optional (?pwd=)Requires extra setup + usually Marketplace approval
Meeting link extraction: Zoom specifics:

​
Lifecycle (High-Level)

Typical flow:
  1. You call POST /bots
  2. Bot joins the meeting (joining, sometimes awaiting_admission)
  3. Bot becomes active
  4. You stop it (or the meeting ends)
  5. Meeting becomes completed and post-meeting artifacts finalize (recordings + final transcript)
You can observe lifecycle changes via:

​
Live vs Post-Meeting

  • Live: use WebSockets to stream segments with low latency
  • Post-meeting: use GET /transcripts/{platform}/{native_meeting_id} for the finalized result
  • Playback (when recording exists): stream bytes from /recordings/.../raw and highlight segments using start_time/end_time
References:

​
Deleting / Anonymizing

DELETE /meetings/{platform}/{native_meeting_id} is designed to be safe to retry (idempotent) and scrubs meeting data for telemetry/usage tracking.