AI Governor.
Independent adversarial validation methodology for clinical AI outputs. Not a medical device. Does not require FDA clearance. Purpose-built to support FDA-regulated AI vendors and the institutions that deploy them.
What it does.
AI Governor compares clinical AI predictions against established peer-reviewed clinical scoring instruments appropriate to the condition under evaluation. Each comparison is cryptographically attested: input hash, scoring-tool version, model output, deterministic output, divergence, timestamp. Receipts are auditable end-to-end.
The point is to disprove the AI on cases where the deterministic tool already gives a defensible answer. When the AI agrees, that is one kind of evidence. When the AI diverges, that is a more useful kind of evidence, and the receipt makes the divergence reviewable by a clinician who was not part of building the model.
What it is not.
AI Governor is not a medical device. It does not produce treatment recommendations. It does not replace clinical judgment, peer review, FDA submission documentation, or any other regulatory obligation that already attaches to clinical AI. It is a validation layer, the kind of independent scrutiny that should exist between a model and the patient it affects.
Methodological foundation.
The methodology is grounded in a pre-registered falsification study published on medRxiv in 2026. 286,510 ICU patients across MIMIC-IV and eICU-CRD. Hypothesis locked on OSF before any data access. Null result on the primary cohort, published anyway. Methodology reviewed by an established sepsis informatics researcher in connection with the credentialed MIMIC access this study required. Read the study.
The discipline that survived that study, pre-register, lock, publish the result regardless of direction, is the same discipline AI Governor applies to every comparison it produces.
Current status.
v1, 21 active modalities. 19 deterministic clinical scoring specialists plus two vision pipelines (chest X-ray, head CT). In active engagement with named pilot partners under NDA. None of these relationships can be named publicly until the parties involved consent.
How it pairs with FDA pathways.
For AI vendors on the FDA SaMD pathway, AI Governor produces the kind of independent adversarial evidence that submission reviewers recognize. For institutions deploying clinical AI, it produces an audit trail that survives medical-device incident review. For FDA-pathway consultants, it provides a peer-reviewable scoring foundation that does not depend on the vendor's own validation framing.
Pricing.
Pricing is set per engagement and depends on scoring-tool count, integration scope, attestation throughput, and pilot duration. We do not publish pricing tiers. Pricing is available under NDA after qualified discussion.
Qualified inbound only.
No public pricing. Use the form below or email Adam directly at [email protected]. Tell us what you are trying to validate and the scope you would consider piloting.