Three key Industrial AI insights for U.S. manufacturing

I recently spoke at the Federal Reserve Bank of Chicago’s Automotive Insights Symposium and started my session with a simple question:

“How many of you have talked about AI in the last couple days--how to institutionalize it or how to apply it?”

Almost every hand shot straight up. That reaction speaks volumes about where manufacturing is today. AI is everywhere in the conversation, but what’s missing is clarity on how to make it real on the factory floor. The excitement is real, and so is the uncertainty.

Manufacturers are trying to understand what AI means for their operations, their workforce, and the systems already running their factories.

In seeking that understanding, there are three key things I hear most often when manufacturers talk about AI. Thes are the critical points of this increasingly important AI conversation—and we’re going to learn a lot by talking with each other.

First: Understanding the problem that you need to solve using AI

No matter the industry, companies want to jump straight to the technology. They want to know what they can do with it and what it can do for them.

I completely understand. AI, digital twins, and automation are exciting. But the very first place we need to go is much less flashy:

First and foremost, we want to understand the problem that you’re trying to solve. And then we want to understand the process. Factories aren’t always greenfield environments. Machines are running today. People are maintaining productivity today. You don’t increase value by ignoring that reality.

So when we talk about future factories or adaptive manufacturing, the conversation always starts with:

• How are you doing things today?
• Where are you losing time, quality, or flexibility?
• What problem actually matters to the business?

Only then does the technology discussion make sense.

This is also where connecting the real and digital worlds becomes essential. Digital twins allow manufacturers to model processes and validate changes virtually before touching the physical environment. Engineering and operations teams can explore improvements with far less risk because the digital environment reflects how the real process behaves.

The digital twin doesn’t replace the process. It helps optimize it.

Second: AI works best when it understands the whole factory

Manufacturers aren’t hurting for dashboards, but they’re starving for insights.

That’s not, however, an AI problem. That’s a context problem. A smart manufacturing survey found that 70 percent of respondents said they were data-rich, but the number-one blocker to operational progress was data quality. I hear the same message across pharma, CPG, and automotive. Although these industries differ drastically, the data challenge remains the same.

When manufacturers talk about AI on the shop floor, they often say:

“I want to walk up to a machine and ask: ‘What was my production today? Why was it down 10 percent?’” AI only works optimally when it understands how all the pieces of a factory fit together. Machines, processes, and production flows are connected in a chain of cause and effect. A motor powers a drive, the drive moves a robot, the robot supports a production line, and the line contributes to overall plant output.

When those relationships are mapped and contextualized, AI can interpret operational behavior instead of just reporting raw numbers. Dashboards can show you what happened, but context shows you why it happened. That’s what enables actionable decisions.

Third: Orchestration is the next-level advantage

Most factories today are a mix of generations, with modern software layered on top of decades of automation, equipment from different vendors, and processes refined over years. Replacing everything is not realistic. The real opportunity lies in orchestrating what already exists.

Industrial intelligence has reached a turning point. Analytics, machine learning, and AI are no longer confined to offline analysis. They are active during operations, predicting maintenance, optimizing throughput, and proposing adjustments in real time

But as intelligence scales, so does complexity. Multiple systems, including scheduling tools, optimization engines, predictive models, and operator support applications, often run simultaneously. Individually they perform well, but without coordination they can conflict, creating instability and forcing people to resolve issues in real time.

The result is not too much automation. It is automation without coordination.

Orchestration solves this. Acting as a governing layer, it aligns intelligent systems during live operations, ensuring actions remain consistent with operational constraints. It allows manufacturers to innovate with AI while relying on proven industrial models to maintain safety, stability, and discipline.

Putting AI into action enables innovation

By focusing on the problem, adding context to data, and orchestrating intelligent systems, manufacturers can move beyond AI hype and turn it into real operational impact. The companies that get this right will not only optimize performance but also build a foundation for the next wave of industrial innovation.

Visit Siemens to see how we're helping manufacturers bring AI into operations.

Published: March 20, 2026