Scribe

Scribe turns recordings into searchable, structured knowledge. Upload audio or video — or paste a YouTube link — and Scribe transcribes it with speaker identification, generates summaries, and makes everything searchable.

The transcript engine produces timestamped, speaker-labeled text. Search works across all conversations, so finding what someone said about a topic is a single query rather than scrubbing through hours of audio.

Built-in knowledge management ties transcripts to people, companies, and tasks. Scribe extracts action items from conversations and tracks them. People and company profiles aggregate mentions across all transcripts.

Documents and contexts provide additional structure — attach reference material and set context for how transcripts should be interpreted.

Export transcripts as markdown with summaries and timestamps. The CLI provides full access to everything — search, transcript inspection, uploads, speaker assignment, task management, people/company lookup, and raw API access.

Built on the Happenings performance stack: Go backend with ConnectRPC, Protocol Buffers, PostgreSQL with sqlc, Next.js dashboard, deployed on Cloud Run.