DemoPricing

What is AI Rule Contradiction Detection?

A contradiction is two or more rules that can apply to the same case and lead to incompatible outcomes. Contradiction detection finds them before the rules are sealed, so a system never has to guess which one wins.

What Studio looks for

Rules that can apply to the same case but ask for different decisions. Studio finds every such case exactly, shows one, and says which rule the sealed policy applies there. Where the policy settles it (a higher priority, a stricter rule winning a tie, a narrower exception), the overlap is counted, not flagged. A person decides the rest: an approval that overrides a stricter rule, a rule that never takes effect, a lower-authority document applied over a higher one, or rule order alone deciding. Studio also flags vague terms such as "reasonable", and differences between the written rules and past decisions.

Why it matters

A rule set that contradicts itself still gives an answer, but not one anyone chose. Studio shows each conflict with a case it happens in, and a normal seal needs a person's decision on each one.

Example

Scenario

A KYC policy says "approve if employment is verified" and "escalate if self-employed income is above 50,000".

Without governance

A self-employed customer with verified employment and income of 55,000 meets both rules. Nobody sees that the policy says two things about this customer until someone asks which rule decided.

With NOMOS

Studio shows the case and that the sealed policy escalates it: at equal priority a review wins over an approval, so nothing needs deciding. Had the approval been given the higher priority, it would override the review for every such customer, and Studio would ask the owner to confirm that exception.

Frequently asked questions

Can contradictions come from two different documents?+

Yes. When several documents are combined, a regulation and an internal procedure for example, a document that says it replaces the other is recommended first, then the higher document type, and a person confirms each conflict.

Related concepts

See it in practice

Upload a policy, confirm its rules, seal them: any system can then check an action against them.

Open Studio →