Record, transcribe, summarize.
All on your Mac.

Yes. MeetingMemo captures your microphone and the call's audio together, at the system level — both sides of any meeting, and you don't need to be the host. There's no BlackHole, Loopback, or virtual audio driver to set up.

No. MeetingMemo records what your Mac hears, like a local recorder. Nothing joins the meeting, no extra participant appears, and nobody gets a notification.

Your audio is recorded and transcribed entirely on your Mac and is never sent to a MeetingMemo server — recording and transcription work fully offline, and AI summaries do too if you run them on-device with Ollama. The only things that ever leave are optional: iCloud sync (to your own private iCloud) and, if you choose cloud AI, the text transcript — sent to the provider you chose, never to us.

No. We never receive your audio or transcripts, so there's nothing for us to train on — and no "opt out of training" setting to remember. If you use AI summaries, the text goes to the provider you picked, under your own account's terms.

Not necessarily. AI summaries, action items, and chat run two ways: on-device through the free Ollama app — no key, no per-token cost, nothing leaves your Mac — or on your own cloud key (OpenAI, Anthropic, Gemini, and others), where that provider bills you for what you use and only the text transcript is sent, never the audio. Recording and transcription need neither: without AI set up, MeetingMemo still records and keeps the raw transcript on your Mac.

Yes. Install the free Ollama app, pull a model, and MeetingMemo finds it automatically — summaries, action items, and chat then run locally, and nothing leaves your Mac at all. One honest caveat: local models are weaker than the big cloud ones, so pick the largest model your Mac runs well — MeetingMemo recommends one and warns you if your choice is too small for reliable summaries.

Run AI on-device with Ollama — nothing leaves your Mac — or bring your own key from Anthropic (Claude), OpenAI, Google Gemini, and other major providers, or a compatible custom endpoint. With a cloud key, MeetingMemo sends only the text transcript, under your own account; your key stays in your Keychain, and nothing routes through a MeetingMemo server. You can also keep calendar attendee names out of those requests.

It uses Whisper-class on-device engines (WhisperKit and Apple's SpeechTranscriber). On clear speech it's on par with the leading cloud services, across about 99 languages with automatic detection. Heavy accents, crosstalk, and noisy rooms lower accuracy, as with any tool.

It labels your voice as You and everyone else as Others by separating your mic from the call's audio — clean attribution for 1:1s and small calls. It doesn't name each individual person in a large group (there's no full speaker diarization in v1).

Every feature is free for 30 days, then $29/year on the Mac App Store, with Family Sharing for up to 6. It's a subscription, not a one-time purchase — that covers ongoing updates and iCloud sync. There's no permanently free tier.

A Mac with Apple Silicon (M-series) on macOS 26 (Tahoe) or later, plus about 2 GB free for the app and a transcription model. AI summaries need either the free Ollama app (for on-device AI) or your own cloud provider key. It doesn't run on Intel Macs or earlier macOS versions.

Recording laws vary by where you and the other participants are — some places require everyone's consent. You're responsible for following the law where you are, and we recommend always letting people know before you record.

Anytime. Export any meeting as Markdown, text, JSON, or XML, and download the audio — transcripts are also saved as plain Markdown on disk. Your library syncs across your Macs through your own iCloud.