Transcript LOL

Label a transcript's diarization speakerIndex (from list_transcript_speakers) with a named speaker (speakerId from list_speakers). Updates the recording's transcript and every labelled view of it.

assign_speaker

Create a folder in a workspace to organize recordings. Optionally nest it under a parent folder. File recordings into it with move_recordings_to_folder.

create_folder

Create a new insight prompt template in a workspace — useful when a task needs an insight no existing prompt covers. Returns the new prompt's id; pass it to generate_insight to run it against a recording. The prompt is saved to the workspace's library and does NOT run automatically on future transcriptions. Check list_prompts first to avoid creating a duplicate.

create_prompt

Create a new named speaker in a workspace. Use this while labelling a transcript when the person isn't already in list_speakers. Returns the new speaker's id — pass it to assign_speaker to label a diarization speakerIndex with it.

create_speaker

Generate an insight for a recording by running one of the workspace's prompt templates against its transcript. Generation is asynchronous: this returns immediately with the new insight in 'QUEUED' status — poll get_insights and read the item with the returned id until its status is 'IDLE' (finished, content populated) or 'FAILED'. Discover promptIds with list_prompts. This consumes the account's AI-generation credits and may require a paid plan.

generate_insight

Fetch the AI-generated insights already produced for a recording — the curated summaries, action items, key points and other outputs from the workspace's insight prompts. Prefer this over re-summarizing the raw transcript yourself: it reuses the prompts the user configured and the work the product already did, into far less context. Returns each insight's prompt title, generated content, and status. Use list_prompts + generate_insight to produce an insight that doesn't exist yet.

get_insights

Fetch a single recording's metadata, including its transcription status. Use this to check whether a transcript is ready before calling get_transcript. To see the recording's translations, use list_translations.

get_recording

Fetch the transcript of a completed recording. Call this when the user wants to read, summarize, quote, analyze, or answer questions about a recording or meeting. Returns readable plain text by default; use 'srt'/'vtt' for subtitles or 'json' for timestamped paragraphs. Optionally pass startTime/endTime to fetch only a time window (e.g. the minutes around a search hit) instead of the whole transcript.

get_transcript

Fetch a translated transcript, in the same formats as get_transcript ('text' by default, or 'json'/'srt'/'vtt'). Get the translationId from list_translations; the translation must have finished (status 'COMPLETE'). Optionally pass startTime/endTime to fetch only a time window. This returns the translation only — use get_transcript for the original-language transcript.

get_translation

List the folders in a workspace (id, name, parentFolderId). Use this to discover folderIds for the folderIds filter on list_recordings or search_recordings.

list_folders