An audio element on your board can be transcribed, and then worked with like any other text — summarised, quoted, turned into notes, turned into a task list.

"What did we agree in this recording?"
"Pull the quotes about pricing out of this interview."
"Turn this voice memo into a task list."

It has to transcribe before it can answer

The AI cannot hear. A board read gives it the file's address and nothing else, so a question about a recording is answered by transcribing it first. It will never describe what a recording says without doing that — a summary of unheard audio is an invention.

It costs money per minute

Transcription is charged by the minute of audio, unlike most of what the AI does. A given recording is transcribed once and the AI works from the transcript for the rest of the conversation, so follow-up questions about the same file are free.

Practically: put your questions about one recording in the same session rather than coming back to it cold.

Use Cases

• Turn a client call recording into a decisions list on the board
• Pull quotes out of an interview without scrubbing the timeline
• Get a voice memo written up as notes while you are still in the car park
• Summarise a long recording before deciding which part to listen to