Cinchor · accountability records for AI agents

A real decision an AI agent made — verify it yourself.

A support agent auto-approved this refund under a stated policy ceiling. The instant it did, the record's fingerprint was written to the Omne chain, append-only and outside the application's reach. You didn't paste a key and you don't have to take our word for it: change any character below and the proof fails, because the chain already holds what actually happened.

Third-party evidence, not a self-attested log. When the decision was made, Cinchor committed its fingerprint to the Omne chain at that moment, append-only. Because it was written then, neither the operator nor Cinchor can go back, rewrite what the agent did, and still pass. That's the whole difference from an internal audit log you'd have to take on faith: a green result here is checkable against the chain, not our database vouching for itself.

The full trust model, concessions included · Verify it in code · Issue and verify your own record