Most AI notetakers send a bot into your meeting. It shows up in the participant list, needs a per-platform integration, and creates an awkward moment every time.
Amlin doesn't. It records your Mac's own audio, at the operating-system level, in two separate channels:
Your microphone → labeled Me
Your system audio (everything coming out of your speakers) → labeled Them
Because the two are captured separately, Amlin knows who said what without any speaker-identification guesswork.
What this means in practice
It works with anything that makes sound: Zoom, Google Meet, Microsoft Teams, Webex, phone and FaceTime calls, webinars, YouTube, and in-person conversations picked up by your mic.
No integration to set up. There's no calendar bot to invite, no meeting-platform connection, no "allow recording" prompt for other participants.
No video, ever. Amlin never records or reads your screen. macOS puts system-audio capture behind the screen-recording permission, and Chromium requires a video track to be present before it will route audio, so Amlin creates one and discards every frame.
Audio is never written to disk. It streams straight to transcription and is discarded. Only the resulting text is stored.
Headphones vs. speakers
Amlin deliberately leaves echo cancellation off, so nothing is stripped out of your own audio. With open speakers, your microphone will also pick up the other side of the call, which shows up as some bleed-through on the Me channel. Headphones eliminate it entirely and are worth using if you care about clean speaker labels.
A note on consent
Recording laws vary by jurisdiction, and some are two-party consent. Amlin doesn't announce itself in the meeting, so telling participants you're taking AI-assisted notes is up to you. Many people simply say it once at the top of the call.