A guided ~40-minute governance diagnostic for any company with an AI assistant in front of customers. It maps what your bot is allowed to say and do, how far it has drifted from your published policies, and whether anyone is actually supervising it — and turns that into a dossier you can put in front of a committee.
A tribunal held a major airline responsible for a refund policy its own chatbot invented. The principle now on the record: the company owns what its bot says.
North American tribunal ruling, 2024The EU AI Act's Article 50 takes effect: chatbots must disclose they're AI, in the conversation itself — and it reaches any business selling to EU customers. Fines up to €15M or 3% of turnover.
EU AI Act, Article 50Guardrails and safety controls now ship free on major bot platforms. "The tools existed and weren't configured" turns an accident into negligence.
Platform safety tooling, 2025+Most companies can't answer three questions about their own assistant: what exactly is it allowed to promise, when was it last updated after a policy change, and who reads what it says to customers. This consultation gets those answers on the record.
A calm, structured interview about your assistant — no judgment mid-consultation, no trick questions. "I don't know" is a finding, never a failure.
A risk level from I to IV computed from your answers by a scored methodology — modeled on established public-sector impact-assessment methodology.
Written for two readers at once: the person who answered, and the committee that has never seen the system. Every finding quotes its basis.
Four levels across financial exposure, legal exposure, and robustness of control. A well-governed bot scores Level I — credibility means knowing how to say "this is clean".
Governed. Controls in place and verifiable. Keep going.
Supervised gaps. Partial or informal controls — a short list to close.
Exposed. Real exposure running unwatched — conditions before continuing.
Uncontrolled. Restrict the bot until the named conditions are met.
One page of facts: what the bot does, who built and configured it, what it may say and do on its own — the inventory most companies have never written down.
Every finding quotes the answer it's built on. Nothing rests on our opinion, which is exactly what makes it presentable to a committee or an insurer.
Every genuine control you already have, documented — a dossier that only lists faults reads as a sales document. This one doesn't.
The obligations that apply at your level, each marked in place / gap / absent — and a clear call: go, go with conditions, or restrict until fixed, with owners and dates.
Yes. The founding tribunal ruling that held a company to its chatbot’s invented policy came from North America, not Europe. The EU rule matters the moment any of your customers are in the EU; the governance questions matter everywhere your bot talks to customers.
Whoever answers for the assistant — a founder, ops or support lead, or the project owner. You don't need documents: the consultation works from what you know, and what you don't know becomes an open item with an owner and a date.
Especially then. "The agency handles it" is one of the most common findings — the consultation establishes what's actually in the contract, who owns the bot's mistakes, and what you can see and change yourself.
A structured document that states what an AI system is, what authority it has been delegated, what risk level that represents, and what obligations follow — in a form a governance committee, board, or insurer can read without having seen the system. AIGovCheck produces one in a single guided consultation.
Private beta — free, by access code, sent personally by email. No commitment, no sales call.
Leave your name and email — we'll take it from there.
Already have a code? Start your consultation from the home page →
Your access code will arrive by email if there's a fit. In the meantime, the scripted demo shows how a consultation works.