AI for manufacturers. Qualify it like any other process.
Somebody in your shop is already using AI. Where it's used makes all the difference: fine in a brainstorm, costly in a quote, and hard to undo in a CAPA. I treat it like any new process: find the failures that matter for how you use it, put controls in place, and prove the controls hold on your own data.
For businesses concerned about how AI might be used in their organization, and which of those uses need controlling before they cost something.Portland Metro and Pacific Northwest in person, remote everywhere else
Find where AI can hurt you. Control it there.
You'd never put a new supplier on the line without a first article. Most shops just did exactly that with AI. You cannot imagine every way your people might use it, but you can see the tasks that matter most, and those are the ones worth ranking. Rank them, decide what level of control each one needs, and put that control where the risk actually is:
Where AI is used in your business, ranked by risk
The failures that matter there, and the controls that fit them
"Someone will notice" is NOT a control your business should run on
Tailored to your processes and your use cases, then a control plan your team owns
What I did with the same discipline
I ran a $65M contract manufacturing plant through a two year turnaround. The money figures are the controller's numbers, not my estimates. This is the operating record behind the method; it is not AI work.
The Root Cause Analyzer
Root cause investigations, from a fast, defensible root cause summary to CAPA and full 8D records. The same investigation underneath either way, gated on evidence, with what a reviewer actually checks: causes verified, alternatives ruled out in the record, and it will not let you stop at operator error. Run a full investigation free; your experts make the call. Built on 30 years of quality engineering. It is also the proof. The same work I would do in your shop, running in public where you can go break it.
Who you'd be working with
Richard Snyder. Thirty years in manufacturing, most of it in FDA regulated medical device plants. I have been the quality manager, the operations manager, and the manufacturing engineering manager, which is why your problem does not have to stay inside one department for me to be useful on it.
"It is not necessary to change. Survival is not mandatory." (W. Edwards Deming)About Richard Snyder