Hard hats and a Responsible AI rollout.

Because the program was built next to the people who would use it—same hard hats, same noise, same shift change. Rolling out from a conference room would miss how operators actually work. Building on the floor let the AI solutions be tailored to the operators' real environment rather than an idealized lab setting.

The model flags a likely defect; the operator decides. That order matters, and the team did not let it drift the other way. Every flag is a suggestion routed to the person who knows the machine—the human has the last word. A defect model is only as honest as the line worker willing to tell you when it's wrong.

No headcount came off the floor. The program moved people up the value chain—into review, calibration, and exception handling—rather than replacing them. This is the workforce-first principle in practice: AI augments the operator instead of displacing them, and the operators' years of pattern recognition were treated as a valuable dataset.

The operators trained the reviewers, because years of pattern recognition is a dataset and the team treated it like one. Their hard-won knowledge of the machines and defects fed the program rather than being sidelined by it. This reflects the rollout's core stance that the human who knows the machine remains central.

Deciding who can change a threshold, who signs off when the model gets retrained, and what happens on the floor when the suggestion screen goes dark. The lesson carried out of the plant is that Responsible AI on a manufacturing floor is not a screen or a model—it's a set of agreements about who decides, written down before anything ships and honored when the screen goes dark.

It means the AI flags and the human decides, in that fixed order. Flags are suggestions routed to the person who knows the machine, no jobs were removed from the floor, and operators moved into higher-value review, calibration, and exception handling. The system's honesty even depends on operators being willing to say when a flag is wrong—so the human stays essential, not optional.