Wispr Flow is preparing to launch a meeting notetaker for Mac, according to updated Terms of Service published July 25 and an email to customers, as reported by TechCrunch on August 5, 2026. The product, branded "Notetaker," uses system audio to transcribe meetings without joining them as a bot participant, then cleans up transcripts, generates action items, and lets users query across meeting history with AI TechCrunch.

The company's Terms of Service define Notetaker as Wispr's AI-powered meeting transcription, note-taking, summarization, speaker identification, and action item generation. The updated terms specify that Notetaker input may include meeting audio, participant information, meeting metadata, speaker labels, meeting transcripts, and other meeting-processed information. Output may include AI-generated meeting transcripts, summaries, action items, meeting insights, speaker attribution, and other meeting-related content TechCrunch. The privacy policy, also updated July 25, states that Wispr processes Meeting Data to generate those outputs.

Wispr Flow's Notetaker product page details how the tool handles speaker identification. It identifies speakers using the user's calendar invite, a personal dictionary, and conversational context, such as when someone says "thanks, Priya." During a meeting, Notetaker provides a live transcript that separates the user from other speakers. After the meeting, it produces a polished transcript with full name labels. Speaker labels are most accurate on video calls where each participant joins from their own device. In shared conference rooms, Notetaker relies on conversation context to identify people when they are mentioned. If a speaker is unlabeled or mislabeled, the user can assign the name once and the correction propagates across the entire transcript Wispr Flow.

The notetaker also provides a live transcript view during the meeting and lets users ask AI to catch them up on what they missed. Beyond the meeting itself, Wispr Flow's MCP documentation, published August 4, describes connecting Flow to AI tools like Claude and ChatGPT, enabling queries such as "What were the action items from my meetings this week?" Wispr Flow Docs.

The launch extends Wispr Flow's trajectory from AI-powered dictation into meeting intelligence. The company released an iOS dictation app on June 3, 2025, followed by a Series A round of $30 million from Menlo Ventures later that month. In November 2025, Wispr secured $25 million from Notable Capital as the dictation app gained traction. An Android app launched February 23, 2026, supporting translation in over 100 languages and cross-app functionality TechCrunch. Wispr Flow's free tier allows up to 2,000 words per week, with an unlimited plan at $12 per month or $144 per year TechCrunch.

In total, Wispr Flow has raised more than $81 million and was valued at $700 million at its last round. Bloomberg reported in May 2026 that the company was in talks for a new funding round valuing it at $2 billion TechCrunch. Co-founder Tanay Kothari had previously discussed building an AI assistant, and the Notetaker product appears to be the realization of that vision.

The system-audio approach distinguishes Notetaker from the dominant pattern in AI meeting tools, where a bot joins the call as a visible participant. That model, used by Otter, Fireflies, Read, and others, has known friction points: participants sometimes object to a bot in the room, IT policies may block external bot joins, and the bot can draw unwanted attention to the fact that a meeting is being recorded. Capturing system audio locally on the user's Mac sidesteps those issues entirely, at the cost of capturing only what the user's machine can hear, which may affect speaker separation in shared-room scenarios.

The speaker-attribution approach is worth examining. By combining calendar metadata, a personal dictionary, and conversational cues, Wispr Flow is attempting probabilistic identification rather than voice fingerprinting. The product page is transparent about the limitations, acknowledging that shared conference rooms reduce accuracy. The one-touch correction that propagates across a transcript is a pragmatic UX choice that addresses the inevitable mislabeling without requiring the user to fix every instance manually.

The MCP integration is a notable signal about where Wispr Flow sees the product heading. Meeting transcripts are only useful if they can be queried, and wiring Notetaker into Claude and ChatGPT via the Model Context Protocol positions meeting data as an input to broader AI workflows rather than a standalone artifact. Whether enterprises will be comfortable routing meeting audio and transcripts through a startup's processing pipeline is a separate question from whether the technology works.

Wispr Flow is entering a crowded category. Otter, Fireflies.ai, Read AI, tl;dv, and Microsoft's own Copilot in Teams all offer some combination of transcription, summarization, and action item extraction. The differentiation Wispr Flow is betting on is the no-bot approach combined with cross-meeting AI querying and the company's existing dictation footprint. The $2 billion valuation reportedly under discussion suggests investors are pricing in the broader assistant vision, not just a notetaker.

The product has not launched yet. What is confirmed is updated terms, a product page, MCP documentation, and a customer email signaling that release is imminent.