Who's behind the work.
Noetika Labs independently measures whether the rules an AI system is given actually hold. Founded by John Gabrick.
John Gabrick
A clinician and technologist who has built and governed enterprise AI, and who built expert systems at Carnegie Mellon.
He is the author of The Reviewer Who Wasn't There, a controlled study of what happens when an AI is required to name who approved its work.
The idea was right. The machinery wasn't.
Giving a machine an expert's rules is not a new idea. It was tried decades ago, back when you could capture the rules but the engine underneath them was brittle — every piece of knowledge hand-fed by specialists who cost more than the answer was worth. So the field stalled and waited for an engine.
The large language model is the engine that finally arrived. Everything that made rule-following machines impractical then is now the cheap part. What was hard is easy. And what everyone assumed was easy — knowing whether the rules are the right ones, and whether the machine is actually following them — turns out to be the whole problem.
Proof, not reassurance
Organizations are putting AI to work on records people rely on. Someone wrote the instructions it follows, and someone else signs for what it produces. The person who signs rarely has a way to know which of those instructions are doing anything.
We find out. The result is a clear answer for each rule — whether it holds, whether it matters, and what to do about the ones that don't — so the person accountable for the output can stand behind it.
What every engagement is held to
We measure what we didn't build
The value of an independent check is that it's independent. We don't write your AI's instructions and we don't sell you the system. Your team builds; we prove what it does.
Measured and assumed never get blended
Every finding says plainly what was measured and what wasn't, and a person reads every result before it reaches you. If the honest answer is that you don't need an audit, that's the answer you'll get.