The Control Room for Autonomous Trading Risk.
Clean proposals settle in milliseconds, while outliers demand human approval.
Six evidenced checkpoints.
All six pass and the trade settles in milliseconds. One fails and it is quarantined with the evidence attached.
The trade executes and the sealed artefacts are written to the chain. Median decision time under 200ms. No human involved, and none needed.
The channel never opens. The proposal lands on the Exception Desk with a reason code. A cryptographic failure needs two operators to override; a policy hold needs one.
Operators only ever see what failed.
A triage queue, not a report. Clean settlement collapses into a feed; anything held sits at the top with the failing check named.
$4.2B settled today
Straight-through
Asset limit breach
AI agent tried to buy an unauthorized asset class.
Triage first
Held trades sit above everything else, tagged proof failure or policy hold.
Kill switch
Halt at platform, institution, agent or asset scope. Effective on the next check.
Headroom live
Remaining daily envelope per institution, so a limits hold is visible before it happens.
A ledger that cannot be quietly edited.
Each entry is hashed with the digest of the one before it. Change a byte anywhere and every digest after it stops matching.
Evidence_Bundle_Trade_#4912.pdf
Sealed · 6 checks · hash-linked
Append only
The database rejects updates and deletes on the ledger. Records are written by a trigger.
Verifiable on demand
Integrity is recomputed across the whole history and reports the first sequence that fails.
Evidence bundles
Any settled trade exports with its terms, six check results, signatures and chain anchor.
We will run a live proposal against your own risk policy.
Thirty minutes: inject a clean trade, inject a broken proof, and watch the difference land on the desk and on the chain.