Human's own safety

Someone moves a rack. Adds a workstation. Changes the flow of a production area.

Operationally, it might feel like a small change.

But the robots’ protective fields, scanner configurations, and speed zones were validated against the old layout.

That’s where things can get dangerous—not because anyone ignored safety, but because two changes that should have been connected never were.

This is exactly where I think agentic AI gets interesting in manufacturing.

A management-of-change ticket or CAD revision becomes the signal.

An AI coworker can identify which robots, zones, and scanner configurations could be affected. It can pull the relevant parts of the site risk assessment, surface the applicable requirements from ISO 3691-4 and ANSI R15.08, and draft the revalidation checklist.

Then it keeps the change ticket open until a human signs off.

And that last part matters.

The AI does not change a safety configuration.

It doesn’t make the safety decision.

It doesn’t approve the system.

The human owns safety.

The AI’s job is to make sure the right information gets to the right person at the right time... and that a layout change doesn’t quietly create a missed revalidation.

That’s the kind of agentic AI conversation I’m looking forward to having at the Robotics Exchange Summit.

Not replacing the safety process.

Making it harder for the process to miss something.