Threadfork

Threadfork is a privacy-first AI meeting notetaker built exclusively for macOS on Apple Silicon. Unlike most competitors in the meeting-intelligence space, it runs entirely on-device: local models handle transcription, speaker diarization, summarization, and entity extraction without shipping audio or transcripts to a cloud backend, which makes it suitable for regulated industries and confidential conversations. The app captures both system audio (Zoom, Google Meet, Microsoft Teams) and microphone input for in-person meetings, then converts each conversation into a structured artifact with summary, decisions, questions, and commitments. Its distinguishing feature is the 'thread' abstraction: commitments extracted from a meeting become actionable items with owners and deadlines, and people, companies and topics accumulate timelines that stitch together across every meeting they appear in, giving you a queryable memory of your working relationships. Semantic search runs over the full transcript and summary corpus so you can retrieve past discussions by meaning rather than exact wording. It works offline, which matters when you're on a plane or in a SCIF-adjacent environment, and it avoids the ongoing subprocessor sprawl that comes with sending audio to third-party ASR and LLM APIs. Typical users are founders, consultants, sales AEs, product managers and lawyers who take enough meetings that manual notetaking has broken down, but who cannot or will not put client audio through Otter, Fireflies or Fathom. It is Mac-only and single-user in its Pro tier — teams need the Enterprise plan.