Every step an agent takes, captured and chained as it happens.
Autonomous agents already decide who gets an account, who gets credit, and which payment goes out the door. The question a regulator asks six months later is not whether the agent was clever — it is what it saw, what it decided, and whether that account of events can be trusted. Press play and watch a real bank workflow run: each agent action becomes a record, each record is hashed onto the one before it, and the chain seals at the end.
Simulated agents on fictional case data. The chain mechanics on screen — SHA-256 linking, Merkle root, offline verification — are the ones Coriqo runs in production.
Three places an agent chain already decides money.
Capture, chain, verify — three things, in that order.
The demo above is deliberately literal. Nothing is summarised after the fact; the record is written at the moment the agent acts, which is the only moment the evidence is still complete. Scale that across every agent in production and this is what a control tower over agent decisions looks like — not a summary dashboard, the underlying record itself.
Written at the moment of decision
Each agent hop — the tool it called, the score it got, the threshold it crossed, the human it escalated to — is recorded as it happens, with its inputs and its actor. Reconstructing this from logs a quarter later is how organizations end up with an explanation instead of evidence.
Each record hashed onto the last
Every block carries the SHA-256 of the block before it. Edit any earlier step — soften a risk score, delete an override — and every hash after it changes. Tampering does not need to be caught by a reviewer; it fails arithmetic.
Checkable without trusting us
The sealed chain produces a Merkle root and an Ed25519 signature. Your examiner re-computes both offline against a public key, from an export they hold. Coriqo is not in the trust path — that is the entire point.
An agent that cannot show its work is not an efficiency gain. It is an unpriced liability that compounds every time it runs.
What the examiner runs, on their own machine.
The seal at the end of the run isn't a badge we render. It is a Merkle root over the chain plus a detached signature, exported as JSON. Anyone holding that export and the public key can check it without an account, without network access, and without asking us anything.
Note: record counts and the anchor index above are from the demo case. Chain, Merkle and signature verification are implemented in the Coriqo codebase today; public anchoring degrades to “pending, retry” without outbound network access.
If you are putting agents anywhere near onboarding, credit, or payments, the interesting conversation is not this demo — it is your actual chain: which hops you would capture, who signs off on an exception, and what an examiner would be handed twelve months from now.
We are working with a small number of design partners on exactly that. Bring a workflow and we will map it.