The best time to capture case details is rarely when you are sitting at a desk. It is in the car after a hearing, walking between buildings, or on a call from a client who finally remembered the name of the witness. CouncilGPT's voice intake is built for that reality.
Talk naturally
A lawyer opens the mobile app and records a voice memo exactly as they would to a colleague. The recording is transcribed with speaker diarisation when more than one person is present, then passed to the Intake agent. The agent does not merely produce text; it extracts structured entities: parties, dates, deadlines, locations, claims, and open questions.
Each extracted item becomes an intake_item in the matter. The item is linked to the original transcript, so anyone reviewing it later can hear or read the source context. Intake items are tagged by type (deadline, party, fact, task, risk) and surfaced in the matter timeline and workflow planner.
From noise to next steps
The value is not transcription accuracy alone. It is turning an unstructured stream of speech into a set of decisions. A two-minute memo about a client call can produce three deadlines, a new opposing party, a request for documents, and a suggested follow-up email. Those suggestions appear as draft intake items; a lawyer approves, edits, or rejects each one before it is committed to the matter record.
Privilege-aware from the start
Voice intake inherits the same confidentiality controls as the rest of the platform. Transcripts are stored in the matter vault, not shared with transcription services beyond what is required for the request, and never used to train models. For lawyers who live by the details they can never quite capture in the moment, voice intake turns memory into structure.