The first release.
The line: audio never leaves the Mac. Transcript text leaves only if you choose your own OpenAI or Anthropic key for notes. No analytics, no account.
- Records the meeting app and your microphone, for Zoom, Microsoft Teams, Google Meet, FaceTime, Slack and Discord, or anything the Mac can hear with ⌘⇧R
- Transcription on the device with Whisper, in your choice of model size
- Coach: live help during the call, updated every few minutes above the meeting window
- Notes as markdown: summary, decisions, action items and open questions, with the full transcript, written to a folder you choose
- Notes written by a local model through Ollama, your own OpenAI or Anthropic key, or not at all
- Dictation into any app with a held key; audio processed in memory and never written to disk
- The Library: every call searchable on the Mac, with reveal-in-Finder and delete
- A first-launch setup that walks through consent, permissions, the speech model, the notes provider and the export folder, resumable across relaunches
- One-time unlock after three free meetings; App Sandbox, hardened runtime and a privacy manifest
The direct download build, with dictation that types straight into other apps, is being prepared. It will appear here when it ships.