How auto-recording of meetings works

Automatic meeting detection, the manual-start hotkey, "you" vs "them" speaker separation, and where the finished notes end up.

Automatic detection

The app watches, in the background, for call apps running on your computer — Zoom, Microsoft Teams, Slack, Discord, Telegram, WhatsApp, Webex, Skype, and browser-based calls. As soon as one of them grabs the microphone, you'll see a notification: "Looks like you're in a meeting — start recording?" with options "Yes", "Don't ask for this app", and "Always for this app".

A hotkey just in case

If your call service isn't on the auto-detect list, or you just want to start or stop recording manually, the Cmd/Ctrl+Shift+R hotkey works at any moment, from any app.

"You" vs "them"

Your voice and the other participants' voices are captured as two separate streams — the meeting's system audio on one side, your microphone on the other. That means the app never has to guess who's speaking: the "you / them" labeling in the finished transcript is accurate from the very first line, with no voice training needed.

Where the notes end up

After the meeting ends, the recording is processed the same way as any other transcription and shows up in the "Meetings" section tagged with source "desktop", as well as in your workspace's general recordings list. Assign it to a project, and the AI will extract entities from it and add them to that project's knowledge base.

Voices of the participants

If regular participants' voices are already confirmed in a project, the AI recognizes them in new recordings too — so the transcript shows real names, not just "you / them" labels. The project's "Voices" tab shows who's already recognized and who's worth naming manually.

While a recording is running, the app's tray or menu-bar icon changes — a quick way to confirm recording is actually on without opening the app window. This still works no matter what hardware the other participants use: the transcript clearly separates you from everyone else on the call.