Will your AI agent survive an audit?
The EU AI Act's high-risk rules are due August 2, 2026 — and whichever date survives Brussels, the evidence requirement doesn't change. If an AI agent touches your claims, contracts, patients, or customers, someone will eventually ask you to prove it's scoped — and a screenshot of a chat is not proof.
What you leave with
A decision boundary
What your agent may decide versus what it may only recommend — written down, in your words, for one real process you own.
A governed spec
Your rules in a form a machine can check. Plain language. No terminal, no code, no engineering degree.
A verdict with evidence
Watch a plausible-looking agent get caught live — then see what a signed, reproducible release record looks like with your name on it.
Who it's for
- VP of Claims, Head of Legal Ops, Clinical Informatics, Head of Support
- Compliance and risk leaders whose name is on the sign-off
- Anyone whose vendor just said "trust us, it's safe"
- Not a coding tutorial — bring an engineer if you like, but you won't need one
- Not a product demo — you leave with artifacts that are useful whether or not we ever speak again
- Not legal advice — we show evidence, your counsel judges it
Save your seat
Seat saved.
We'll email you the date, the calendar invite, and one short pre-read.
Nothing else. No sequence, no sales follow-up.
Why "Assize"?
The medieval assize was history's first statutory quality gate: bread, ale, and cloth were examined against a written standard before they could be sold. Centuries later, the principle holds. An agent should stand examination before it ships — and the examination should leave evidence.