You Automated It. You Didn't Solve It.

You did it.

You got the funding approved, brought in the integrator, and automated the process. The cell is running, the robot is moving exactly the way it was designed to, and parts are coming off the line hitting TAKT time.

It looks like everything worked.

And then rework shows up.

Not all at once—just enough to make people pause. Just enough to introduce doubt into something that was supposed to be solved. And that’s when the question shifts from, “Did we automate it?” to something much more uncomfortable.

Why isn’t it right?

What people don’t expect is how fast the pressure moves. It doesn’t stay with the operator anymore—because there isn’t one—so it lands on maintenance, engineering, and management. And the stakes are higher now, because this isn’t a person with a tool in their hand.

This is an investment that was supposed to fix something.

When I see a plant going down the route of automation, it usually started with frustration over "unreliable tools".

That's when we hear, “We spent all this money. Why aren’t we running?”

Or worse—“We’re running, but we’re sorting parts.”

Day 1 vs. Day 191

The easiest problems to fix are the ones that never work at all. Everything stops, everyone pays attention, and the issue gets solved.

But when something works—consistently—and then doesn’t?

Those are the problems that sting. They move through production, show up downstream, and cost more every time they’re touched.

They don’t stop the line.

They just quietly drag it down.

And the longer it’s been since the robot went online, the more instinctive it becomes to hide it—because now it feels like something you broke.

Most of the time, you didn’t.

Bottom Line It

I’ve seen this play out more times than I can count. A friend of mine—Robot Rob—called me one day looking for a tool recommendation. He told me later he expected I’d give him a model number and move on.

Instead, I asked questions about the application, the substrate, the location in the plant, and how the system was set up. About 90 minutes of windshield time later, he laughed: “Well, I guess I’ve got some homework to do. And here, I thought you were just going to tell me what tool to buy.”

This is a professional automation and robotics guy with over 20 years in this space. He wasn’t being rude. He just didn’t have the answers to the questions I was asking—beyond the part number of the rivet and how many they were setting annually.

And I wasn’t refusing to give him a model number and a price. I knew off the top of my head which of my tools could set that rivet.

But without understanding the full story, I wasn’t going to risk my reputation—or his—by recommending something that might not actually work in the application.

Because I knew something my robot buddy didn’t yet.

This isn’t just a tool decision.

It’s a system.

Quiet Bias

What you’ve been trained on shapes what you see first, and if you’ve spent enough time in one lane, everything starts to look like it belongs there.

And the real problem stays in place.

I’ve walked into plants where it all looked dialed in—good equipment, proper FRL setup, everything maintained—and they’re still struggling. Not because the tool is wrong, but because something simple hasn’t been verified.

A nosepiece. Hydraulic fluid. Something small enough to overlook, but big enough to affect performance.

No one checked. They assumed.

You can have a robot that hits the same position every time, a tool that delivers exactly what it’s supposed to, and a process that looks perfect on paper. But if there is any misalignment in the substrate, none of that matters.

An operator sees that immediately.

They feel it and adjust without thinking—wiggling parts to bring holes into alignment, making it work in real time.

And that’s where you start to see bias show up—operators compensating in real time, doing whatever it takes to make it work… even if that means pre-loading rivets into the substrate just to keep things moving.

A robot doesn’t do that.

It does exactly what it’s told, the same way, every time.

Even if it’s wrong.

Just like a blind rivet tool.

It only knows one job—and it’s going to try to do it every time you pull the trigger.

Quiet Failure

There’s a line Rob says—one I’ve written about before:

“How far, how fast, how heavy—doesn’t matter. It must be repeatable before we talk about anything else.”

That’s the baseline.

But it raises a different question.

In an automated blind riveting work cell, what happens when something starts repeatable—and then quietly stops being repeatable?

That’s when things get missed.

Harder to catch.

Harder to assign blame.

I’m Not Anti-Automation

I support it. But automation doesn’t fix a broken process—it exposes it. And in blind riveting, that matters more than most people realize.

No matter when your robotic work cell started running, the goal isn’t to figure out who’s responsible. The goal is to understand what actually happened.

Because until the system is understood, the problem doesn’t go away.

It just finds a new place to land.

Until next time... M