Vibe Meet Get notified
In development · Windows first

The AI meeting copilot only you can see.

Vibe Meet sits on your desktop, not in your call. Live captions float over any meeting app, a suggested answer is ready when you're put on the spot, and you leave with the recording, transcript and report — saved to a folder you choose, under API keys you own.

No mailing list, no tracking — the button opens an email to the person building it. You'll hear back when there's something to try.

Works with Zoom, Google Meet, Microsoft Teams, Webex, Slack — and anything else that makes sound. Nothing joins the call.

Why it's different

Meeting AI, without the parts you hate.

No bot seat, no "this call is being recorded by a third party" banner with your name on it, no vendor holding your transcripts.

No bot in the participant list

Vibe Meet captures audio on your machine — your mic and your system sound. Nothing dials into the meeting, nobody gets a notification, and the other side installs nothing.

Invisible in screen shares

The caption overlay is excluded from capture on Windows and macOS, so you can present while reading it. Clicks pass straight through it, and it follows your cursor or docks where you put it.

Ctrl + Shift + X — the panic key. One press hides everything and stops audio cold; press again to resume the same meeting.

Your keys, your files

Captions and insights run on API keys you bring, stored in the OS keychain. Every artifact lands in a local folder you picked. No account, no server, no telemetry.

What it does

In the order a meeting actually happens.

From the moment a call starts to the follow-up you write a week later.

1

The call starts

It notices before you do

When a Zoom, Teams, Webex or Slack call goes live — or a Meet tab starts using your mic — a small always-on-top prompt offers to start the copilot. Or it just starts by itself, your pick. (Windows today.)

2

While you talk

Captions, translation, and an answer when you need one

Live captions from four interchangeable engines — OpenAI Realtime, Speechmatics, Groq, or fully on-device — with partials landing in about a second and a half.

Suggested answers. When a question is aimed at you, an answer is already on screen — plus follow-up questions worth asking and a rolling summary that keeps you oriented after you zoned out for ten minutes.

Translation that doesn't patronize. Full live translation when you want it, or smart mode: captions stay in the original language and only rare words get an inline gloss. A local frequency check means most captions never trigger an API call. (English → Portuguese first; more pairs later.)

3

You hang up

Everything lands in one folder

One timestamped folder per meeting: the recording with you on the left channel and everyone else on the right — that split is the speaker separation — a timestamped transcript, and a report: overview, decisions, action items with checkboxes, risks, and paste-ready build prompts for every feature that got discussed.

Saving never depends on the report: if a report call fails, your audio and transcript are already on disk, and you can generate the report later.

4

Days later

Ask the meeting itself

The Meetings tab reads your folder back: browse every meeting as a list or a month calendar, click any transcript line to hear that exact moment, keep notes that autosave, and re-transcribe or generate a report whenever.

And every meeting has a chat. It answers from that meeting's transcript, report and notes — with the real cost of each reply printed under it.

Privacy

Private by architecture, not by policy.

There's no Vibe Meet backend to trust, because there isn't one. Here is the entire data flow:

Audio

Goes to the caption engine you chose, under your own key — or nowhere at all with the on-device engine. Never to us; there is no "us" server.

API keys

Stored in the OS keychain — Windows Credential Manager, macOS Keychain, libsecret. Never written to a config file, never shown back to the UI.

Artifacts

Recording, transcript, report and notes exist in exactly one place: the folder you picked. Delete the folder and they're gone.

Everything else

No account. No telemetry. No analytics SDK. The panic hotkey pauses capture instantly — while it's paused, zero audio leaves the machine.

Running cost

You see every cent, because you're the customer of record.

Vibe Meet runs on your own provider accounts — so the meter on screen is your actual bill, not a markup.

≈ $1/hour

for cloud live captions, billed by your STT provider — typical, not a plan.

$0

with the on-device caption engine and local translation. Insight and report calls cost cents per meeting.

live

cost estimate in the tray during every call, and a lifetime spending ledger in settings.

Pricing for Vibe Meet itself will be announced at launch. While it's in development, bring-your-own-keys is the whole model.

Platforms

One codebase, three desktops.

Tauri + Rust. Where an OS can't support something — stealth, follow-cursor, a global hotkey — the app detects it and says so in the UI instead of breaking quietly.

  • Windowsbuilt and validated first
  • macOSsame codebase, being finished
  • Linuxcompiles, being verified
Questions

Fair questions, straight answers.

Does it join my call as a bot?

No. It runs on your desktop and captures audio locally — your microphone and your system sound. Nothing appears in the participant list, and the people you're talking to don't install or see anything.

Where does my audio actually go?

To the caption engine you picked, under your own API key — OpenAI, Speechmatics or Groq — or nowhere: the on-device engine transcribes locally. Recordings, transcripts and reports are written only to a folder you choose. Vibe Meet has no servers of its own.

Can people tell I'm using it?

The overlay is excluded from screen shares on Windows and macOS, and best-effort on Linux. A panic hotkey hides everything and pauses capture in one press. On a physically shared screen or a camera pointed at your monitor it's visible like anything else — no software can change that.

What does it cost to run?

You pay your AI providers directly: cloud captions run about $1 per hour, insight and report calls cost cents per meeting, and the on-device engines are free. A live meter shows the estimate during the call. Pricing for the app itself will be announced at launch.

Is recording meetings even allowed?

Rules on recording and consent vary by country and company — that part is on you. What Vibe Meet guarantees is control: everything it produces stays in one local folder, so what exists, and what gets deleted, is always your call. Recording is also optional — captions and insights work without it.

Want it the day it ships?

Windows build first, macOS and Linux behind it. One email when it's ready — that's the entire marketing plan.