Vision

The largest unbuilt surface in software is the one you talk to.

Orinda Labs is building the intelligence layer for conversations that happen between real people, in real places — and it can only be built with consent.

Almost every system you use was built on things people typed: queries, forms, messages, clicks. It is an extraordinarily rich record, and it is also a record of people alone with a screen. The moment a person turns to another person and speaks, software stops being able to help.

That is a strange place for the industry to have ended up, because the spoken moment is the higher-signal one. A person who says out loud, to your face, what they are trying to do has told you more in forty seconds than a month of their browsing ever will. And unlike browsing, they knew they were telling you.

Today, the tools built on those moments treat the person as a row to extract. The value accrues to one side and evaporates the moment the event ends.

We think the durable version is two-sided. The person who spoke gets an AI that remembers and acts on what they cared about, wherever they go next. The person who listened — the exhibitor, the host, the organizer — gets an AI that captures the room and follows through. Both sides get something out of the same moment, which is the only arrangement in which both sides keep showing up.

What that produces over time is something no amount of scraping can manufacture: a consented map of real-world intent — expressed in person, owned and controllable by the people in it. Not inferred from behaviour. Not bought from a broker. Given, deliberately, by someone standing in front of you.

Why events first

Events and venues are where we started, and they are not the whole plan. They are the best possible first market for this: the consent is unambiguous because the two people are standing together and one of them is asking; the conversation is unusually high-signal because nobody attends an event by accident; and the follow-through is worth real money to the host, so the software has to earn its place immediately.

Everything we learn making conversation legible in a loud hall — the capture, the consent flow, the understanding, the acting on it — is the same machinery wherever people meet in person. Raplo Capture is one application of it. It is not the shape of the company.

Attendee their own AI Consented moment face to face, with permission Host or exhibitor captures and follows up Nothing enters the loop without the visitor’s explicit consent, and they can withdraw it.
Raplo Capture ships the host side of this loop today. The attendee-side products are in development.

The product line

One product shipped. Five being built.

Raplo Capture is the only product available today. Five more are in development — not purchasable, not on a waitlist, and subject to change. They are listed in full, with what each one does, on the products page.

See it working — try Raplo Capture free