The transcript
The transcript is deliberately not a document — it is read only through the meeting itself.
The transcript is not a document
Once every segment is transcribed and the meeting is ended, a single stitched transcript is built and linked to that meeting — but it does not enter the document pool.
So it appears in no document list, surfaces in no workspace-wide search, and the general assistant does not answer from it. That is not a temporary limitation; it is the decision: the word-for-word record of a room should not be reachable by any route except that room.
The way to read it is the meeting's own page. In that meeting's conversation you can ask about the transcript, and the answer cites the passages it was built from.
Persian transcription quality
Transcription is built for Persian rather than translated from an English-first model. The text improves with the names and terms of that particular meeting.
The names come from the meeting itself: its participants, the people seated at the anchoring organisation, and the anchoring engagement with its service. The whole contacts directory is never loaded in — a big list does not raise accuracy, it raises cost.
If the model stalls and repeats a phrase over and over, that run is collapsed and the text keeps a single occurrence.
The transcript is built once
Stitching only happens when the meeting has ended, every segment has succeeded, and no transcript exists yet.
Running it again does nothing; a second transcript is never created.
Read this next
- MeetingsAsk about a meetingA private conversation about that meeting's transcript — even before it has finished.
- MeetingsWho can see a meetingA meeting is private for its whole life. Access is granted one person at a time.
- DocumentsRestricted documentsWhen a document should exist for a named few and nobody else.
Still stuck?
If the answer wasn't here, write to us. It is a real inbox and it gets answered.