Wispr Flow launches Notetaker, a meeting AI with no bot in the call
By Steven Van ·
The free Mac app transcribes calls via system audio, summarizes by topic, and feeds meeting history to Claude and ChatGPT over MCP.
Wispr Flow built its reputation on killing the keyboard: you talk, it types clean text into any app. Now the company is going after another kind of typing, the frantic note-scribbling during meetings. On August 5 it launched Notetaker, a Mac app that captures, transcribes, and summarizes your calls, with a twist that separates it from most of the category: nothing joins the meeting. The announcement came with a very on-brand line: "It's no secret how we feel about typing at Wispr Flow. The truth is, we hate messy meeting notes too."
No bot in the room
Most meeting assistants send a visible bot participant into your call, which changes the room's dynamics and requires everyone to tolerate the recorder in the participant list. Notetaker instead captures system audio locally on your Mac, so it works across Zoom, Google Meet, Microsoft Teams, Slack Huddles, Discord, and browser-based calls without appearing as an attendee. It even covers in-person conversations through the Mac's microphone, and it can detect back-to-back meetings that never made it onto your calendar. TechCrunch framed the launch as Wispr Flow entering the crowded space occupied by Granola, Fireflies, Otter, Read AI, and Fathom, and the botless, Granola-style approach is its clearest differentiator.
Transcripts that know who said what, and how to spell it
Notetaker identifies speakers by name rather than tagging lines as "Speaker 1," with one-click correction when it guesses wrong. It also leans on context to fix the classic transcription failure of mangled names and product terms: calendar invites, your personal dictionary from the Flow dictation app, and connected Gmail and Slack data all inform the transcript. From there it produces structured summaries organized by topic, with timelines, decisions, and next steps pulled out, plus a live "What did I miss?" catch-up if you zoned out mid-call, and a pre-call briefing that researches external participants before you join.
Your meeting history becomes AI context via MCP
The most forward-looking piece is the Model Context Protocol integration. Notetaker exposes your meeting history to Claude, ChatGPT, Cursor, and other MCP-capable tools, so you can ask your AI assistant questions that draw on months of conversations, with source attribution back to the specific meeting. You can also query across meetings and linked messages and emails inside Notetaker itself. For switchers, there is one-click import of meeting history from Granola.
Mac first, free tier included, audio deleted after processing
Notetaker is Mac-only at launch, with Windows announced as coming and mobile on the roadmap. English is the supported language for now, with more promised. Pricing is friendly: it is included in Wispr Flow's free forever plan, with the Pro plan removing limits and Flow for Teams covering whole organizations. On privacy, Wispr Flow says meeting audio is encrypted, stored only temporarily, and automatically deleted after transcription and verification; the company says it does not create voiceprints, sell content, or train models on conversations without consent, and you can delete meetings at any time. Those commitments matter more than usual for a tool that listens to system audio all day.
For creators, the pitch is straightforward. Client calls, sponsor negotiations, podcast pre-interviews, and team standups all generate decisions that evaporate by the next morning. A notetaker that captures them without an awkward bot in the participant list, then makes them searchable from the AI tools you already use, is a genuinely useful addition to the stack, and the free tier makes it easy to test. Wispr Flow, which has raised more than $81 million at a reported $700 million valuation, is betting the same voice-first crowd that adopted its dictation will hand it their meetings too.
Sources: Wispr Flow Notetaker, TechCrunch, Digital Trends