How we work

How we use AI

The standard we hold ourselves to, and the one we will help you build.

We use AI throughout our own work. A person reviews everything before it reaches you, and a person is accountable for it. That is the standard we will help you build, and we hold ourselves to it first.

Why we say so

A firm selling AI adoption that does not use AI itself is the one with the credibility problem. Our own delivery is the demonstration. If the method does not hold up in our work, it will not hold up in yours.

What accountable means here

Not that the work was reviewed. Review is a process claim, and you have been sold process before. Accountable means a named person released it and answers for it. When something gets through that should not have, we own it and fix it.

That is not a weakness of the method. It is why the method needs a person at the end of it.

On your engagement

We use AI on client work, including yours. Every deliverable is reviewed and released by a person, and that person is accountable for it. The same will be true of whatever we help you build: the tool does the work it is good at, and your people keep the judgment.

What we will not claim

• That AI makes no errors. It does.
• That review catches everything. It does not.
• That the tool decides. It does not, and it should not.

Anyone telling you otherwise has either not run this in a real operation or is hoping you do not ask.

In a regulated plant

If you make product under 21 CFR Part 111, ISO 22716 or a similar standard, the question is not whether a model is trustworthy. It is who approved the output, against which revision, and where the record is. That is a process question with a process answer, and it is the same gate your quality system already runs on.

We do not ask your quality team to lower a standard. We put the work inside the gate they already hold.