Why we're building Meeteor

Every hard call has the same moment. Someone asks a question, and the answer exists: it's in the master services agreement, the proposal you sent last month, or the call where they first raised it. You just can't find it while they're waiting.

Meeting AI has mostly solved the wrong half of that. Notetakers write a beautiful summary after the call is over. Live assistants answer instantly, but from whatever a language model happens to know, which is rarely what's in your contract.

Context is the product

Meeteor starts from the opposite end. Before the call, you give it the project: the documents, the notes, the glossary of names and acronyms. During the call, it listens for the moment a fact matters and shows the passage that answers it, with the page it came from. After the call, it writes down what was decided and promised, and that becomes context for the next one.

We'd rather show you the clause than a confident paraphrase of it.

Proof, not vibes

A confident wrong answer during a negotiation is worse than no answer. So every card that quotes a source is checked: the quote has to actually appear in the document or transcript it cites. If it doesn't, the card loses its source badge and is marked as the model's own view. You always know which is which.

Private by default

Meeteor is a desktop app, not a bot that joins your meeting and not a service that keeps your recordings. You bring your own AI key and pay the provider directly, or you run everything on your own computer. There is no Meeteor account and no telemetry. Our Transparency Center shows every request the app makes.

Where we are

Meeteor grew out of Local Motion, a real-time copilot we built for legal calls. The Windows 11 version is in private beta now; macOS is next. Meeteor Cloud, an end-to-end encrypted way to take your context to every device and share it with your team, is in development.