Notetaker is a 📝Wispr Flow product that records meetings from a computer's system audio without joining the call as a participant, then turns them into cleaned transcripts, speaker-attributed summaries, and extracted action items.
It shipped for Mac on August 5, 2026, twenty-two months after Flow's own macOS debut, and enters a crowded meeting-notes category as a direct answer to 📝Granola — down to a one-click import for anyone switching over. Because capture happens at the audio layer rather than through a meeting bot, it works on Zoom, Google Meet, Microsoft Teams, Slack Huddles, Discord, browser calls, and anything else the microphone picks up.
It runs on Mac in English, with Windows in progress and mobile on the roadmap. Notetaker is included in Flow's free plan under usage limits; unlimited use requires Flow Pro at $15 per month, or $12 per month billed annually.
Key Features
- Passive capture — Records from system audio across every calling platform without ever joining as a visible participant or announcing itself to the room.
- Speaker attribution — Identifies participants by name rather than generic labels, so transcripts read as conversation instead of an undifferentiated block.
- Structured summaries — Extracts decisions, timelines, and next steps organized by topic, plus a "what did I miss" catch-up on the most recent stretch of a call.
- Contextual search — Answers questions across past meetings, emails, and messages, and briefs you on external participants by pulling prior threads before a call.
- MCP integration — Exposes meeting history to Claude, ChatGPT, and Cursor over the Model Context Protocol, and imports an existing Granola archive in one click.
Getting Started
- Install Wispr Flow for Mac and sign in — Notetaker ships inside the same app.
- Enable Notetaker from within Flow and grant the audio permissions it requests.
- Start a call on any platform; Notetaker captures system audio without joining.
- Review the transcript, summary, and action items once the meeting has ended.
- Import an existing Granola archive, or connect an MCP client to query history.