Live answers from your systems, read-only.
Order status, availability, compatibility — the agent can look it up mid-conversation through connectors to your APIs. Read-only by design, logged in every transcript.
Look up, never touch
Connectors are read-only GET requests by design. The agent can check a status; it cannot change, book, cancel or pay — ever.
Every call in the transcript
When the agent consults your system, the transcript shows it — what was asked, what came back, and how it shaped the answer.
Enabled by a human, per endpoint
Each connector and endpoint is switched on explicitly by a person, with credentials encrypted and injected only at request time.
From the console
Questions buyers actually ask
Which systems can it connect to?
Anything with an HTTP API that answers a GET request — booking engines, order systems, custom backends. Each connector defines its endpoints and how responses are shaped into answers; enabling any of them is an explicit human decision, never automatic.
Can the agent modify data in our systems?
No. Connectors are read-only by design — the agent can look up, not act. Writing, booking, cancelling or paying is deliberately outside its reach; those remain human actions in your own systems.
Is this safe to expose?
The connector layer enforces HTTPS only, blocks redirects and private addresses, caps response sizes, and treats whatever your API returns as data — never as instructions. Credentials are encrypted at rest and injected only at the moment of the request.
Who builds the connectors?
Setup is part of onboarding — we wire the endpoints with you, you keep the on/off switch per endpoint. Responses are shaped so the agent reads exactly the fields it needs and nothing more.
Answers your customers can trust.
Grounded in your knowledge, honest about its limits, and handed to your team the moment it should be.