Your past recordings.
Your next decision.

TOM records your meetings, voice notes, and screen feedback, transcribes them as you talk, and answers questions across everything you've recorded, all on your Mac

Compounding value

The longer you use TOM, the smarter it gets.

Every recording adds to a searchable memory on your Mac. The more TOM knows, the more it can connect — decisions, promises, patterns.

01Month 1

Transcripts, summaries and action items — searchable per recording.

02Month 3

Q&A across recordings. Ask once; TOM cites the recordings that answered.

03Month 6

Your full history. Search any promise, decision, or recurring topic across everything.

Features

Stored on your Mac. AI on your Mac.

No bots on your calls

No meeting assistant joins the call — nothing for participants to click “kick out.” TOM just records what's on your screen. Works for Zoom, Meet, Teams, Webex, Slack huddles, even an in-person chat on speakerphone.

Video + audio of every call

TOM saves the whole meeting — what was on screen, what was said by you, and what was said by the other side — as a normal video file on your Mac. Go back and re-watch the slide, the demo, or the moment they said the thing.

Live transcript

TOM transcribes while you talk, word by word, with an on-device model. The transcript is ready when you stop. Turn on Whisper refinement in Settings for a slower, more accurate second pass.

Screen feedback for your coding agent

Record a window or your screen while you talk through what should change. Press ⌥A to draw on the screen, ⌥M to mark a moment. TOM pairs every mark with what you said, then copies the screenshots, notes, and full transcript so one paste hands them to Claude or ChatGPT.

All AI runs on your Mac

Transcription, summaries, and search all run on-device on Apple Silicon. After the first launch, TOM works offline indefinitely. No cloud round-trip, no third-party AI provider in the loop.

Title, summary & action items

After every recording, TOM writes a clean title, a TL;DR, decisions, discussion topics, action items, and open questions. Don't like the structure? Edit the prompts in Settings → Advanced.

Search everything you recorded

Ask a real question — “what did we agree with Acme on pricing?” — and TOM finds the answer across all your calls and notes, with citations and the exact moment to re-watch. ~50 ms across a thousand recordings, 100% recall, all local.

Speaks 100+ languages

Multilingual embeddings and a multilingual transcription model mean non-English recordings stay searchable, summarisable, and side-by-side with the rest of your library.

Karaoke playback

Re-watch any recording with synchronised subtitles that highlight word-by-word as the speaker says them. Find the moment by reading, not scrubbing.

Voice notes

Just hit record on yourself for the thinking-out-loud moments — walk, drive, whatever. Goes through the same transcript, summary, and search as everything else.

Drop in old recordings

Got a recording from somewhere else (Zoom export, voice memo, an old MP4)? Drop the file into TOM and it's treated like any other recording — transcript, summary, searchable.

Export any recording

Save a recording as Markdown, an MP4 video, or M4A audio. Feedback sessions also export as a spec folder with every screenshot, ready for a coding agent.

How it works

Four steps, one menu bar icon.

01

Click TOM in your menu bar

Pick a meeting window, your whole screen, or just your voice. Zoom, Meet, Teams, Slack huddles, a voice note on a walk, or feedback on the app you're building — if it's on screen or in the room, TOM can capture it.

02

TOM records on your Mac

TOM captures the screen and the audio from both sides of the call and transcribes it live. No bot joins the meeting. Just remember to tell the others you're recording.

03

Notes ready when you stop

On-device AI turns the audio into a transcript and writes a clean summary, topics, and action items. No cloud round-trip — all inference runs locally on Apple Silicon.

04

Ask questions across everything you recorded

Ask anything. TOM answers from your full history with the exact moments to re-watch — no more hunting through old recordings.

Privacy

Local-first, all the way down.

Your recordings stay on your Mac. The AI runs on your Mac too.

Three jobs need an AI: turning audio into text, writing the summary and action items, and answering questions across past recordings. TOM does all three on-device, on Apple Silicon — no third-party AI provider is involved. After the first-launch model download, TOM works offline indefinitely.

01
Recording — happens on your Mac.TOM saves the video and audio in a folder on your Mac. Nothing uploaded.
02
Transcript & summary — happen on your Mac.On-device transcription writes the words; a local LLM writes the title, summary, action items, decisions, and open questions. Both run on Apple Silicon. No audio or text leaves your machine.
03
Search across recordings — happens on your Mac.TOM embeds every recording locally with a multilingual model and stores the vectors next to the transcripts. When you ask a question, the embedding, the matching, and the cited answer are all generated on-device.
04
We — never see your data.There is no server in the loop. No account, no telemetry, no analytics, no sync. The only network call TOM ever makes is the one-time AI model download on first launch.
No cloud AI provider, no server. No telemetry, no analytics, no account. Your recordings, transcripts, embeddings, and summaries never leave your Mac.

Free. Forever.

Recording, transcribing, summarising, importing, and searching across all your recordings — all on your Mac, no cloud round-trip. Every feature, no licence, no card. The source is on GitHub for anyone to read, build, and contribute to.

Freeforever
  • Free forever — no card, no account
  • Unlimited recordings
  • Every feature included — no tiers, no add-ons, no upsells
  • Fully on-device — your data never leaves your Mac
  • Open source — read, build, and contribute on GitHub
Download for MacView source on GitHub

No catch: TOM is free and open source. Everything runs locally on your Mac — playback, transcripts, summaries, action items, recording, importing, and cross-meeting search. Nothing to renew, nothing to unlock.

FAQ

Questions, answered plainly.

Do I need to bring an OpenAI key or sign up for anything?+
No. There is no account, no API key, no third-party AI provider in the loop. TOM does transcription, summaries, and search across your recordings entirely on-device on Apple Silicon. The only network call TOM ever makes is the one-time AI model download from Hugging Face on first launch (about 4 GB). After that you can use TOM offline indefinitely.
Which AI models does TOM use, and where do they live?+
Can I edit how TOM writes summaries and titles?+
Does TOM identify speakers by name?+
Does TOM transcribe live while I record?+
Do I need consent to record meetings?+
Will other participants see TOM on the call?+
How fast is search across a large library?+
macOS only? What about iPhone or Windows?+
How much disk space does TOM use?+
Can I export my data?+

It's time to put AI to work.

Free and open source, no card or account required. All AI runs on your Mac — no cloud, no key. Whatever you record stays readable on your Mac forever.

Created by
Mircea Preotu