When AI Goes to Work - Transforming Factory Automation

Episode Summary

Industrial AI is moving from the lab to the factory floor. In this episode of Optimistic Outlook In Five, Lauren Espin explores how AI can help manufacturers address skilled labor shortages and improve engineering productivity.

At the center of the story is Eigen Engineering Agent, developed by Siemens’ AI hub in Seattle. By combining AI with plant-specific engineering data, Eigen can help automation engineers write code and configure industrial systems—performing complex engineering tasks at up to five times human speed. In pilot deployments across more than 100 companies in 19 countries, Eigen has delivered up to 50% greater engineering efficiency.

But bringing AI into industrial environments requires more than powerful models. Human expertise, simulation and validation remain essential to ensuring AI-generated work is safe, reliable and ready for the real world.

The result? A glimpse at how Industrial AI could reshape manufacturing—not by replacing engineers, but by helping skilled people do more.

Full Transcript

I'm Lauren Espin, and this is The Optimistic Outlook In Five, a quick update about an important development shaping the future.

The United States has set an ambitious goal to reindustrialize its economy. That means building new manufacturing capacity, modernizing existing operations and remaining globally competitive—all while industry faces persistent shortages of skilled labor.

Industrial AI could be one of the most important tools for meeting that challenge. It can build on decades of progress in automation, software and the digital twin to help engineers, operators and technicians become more productive, accelerate innovation and solve increasingly complex problems.

But realizing that opportunity requires more than powerful AI models. We need to deploy AI in the real world, at scale, in ways that deliver measurable productivity gains.

And that brings us to a piece of technology most of us rarely think about—unless, of course, you work in manufacturing.

The PLC, or programmable logic controller. One may have controlled the line that filled the bottle of water on your desk. Another helped manufacture a component inside the car you drove today. And another may have helped pack the delivery box that showed up on your porch.

PLCs are everywhere in modern industry. Siemens makes a lot of them. In fact, about a third of the machines in the world are controlled by Siemens PLCs.

And every one of those controllers needs to be programmed.

That's the job of automation engineers. They write control code, configure systems and networks, test connections and make sure increasingly complex industrial systems work as intended.

And today, those skilled engineers are in short supply.

So, the obvious question: Why not use AI to augment the productivity of the workforce we have?

AI is already hard at work in factories today. Cameras spot defects on production lines. Systems predict when motors are about to fail. Others find ways to squeeze more throughput out of a critical line.

These applications work because they're really good at doing a specific job. One camera. One production line. Give the AI enough examples, and it learns to do that job extremely well.

But programming a controller isn't one job.

An automation engineer might be asked to solve any number of problems, and the right answer depends on the specific plant—the equipment, the configuration, the wiring, even the naming conventions.

That makes it difficult to scale AI in the same way. You can't build and train a new model for every plant and every customer.

Meanwhile, general-purpose AI solves part of that problem. Today's most powerful models can reason, write software and tackle a much wider range of tasks. But they don't inherently understand your factory. The engineering data that describes a plant—its equipment, configuration and electrical systems—is specific to that environment and often private.

That's where industrial context matters.

Think about a brilliant new engineer on their first day. They understand engineering, but they've never set foot in your plant. You don't send them back to school. You give them access to the right information about the specific system they're working on, and very quickly, they are contributing.

That's the idea behind the Eigen Engineering Agent.

Developed at Siemens' AI hub in Seattle, Eigen works within the engineering software automation engineers already use. It can access the project information that describes how a plant is configured, along with electrical wiring diagrams and other relevant engineering data.

An engineer can describe what they want in plain language. Eigen goes beyond generating suggestions to autonomously executing complex engineering tasks. It can understand the challenge and basically go to work—writing code or configuring systems—until it delivers the performance goal set by the engineer.

And it's already being deployed.

In pilot deployments across more than 100 companies in 19 countries, Eigen delivered up to 50 percent greater engineering efficiency. It can execute engineering tasks at two to five times human speed.

When skilled automation engineers are scarce, those gains matter.

For Industrial AI to be deployed at scale, it also has to be trusted.

That's why human oversight and validation are built into the process. Code can be tested through simulation tools that model how the controller will behave before it reaches live equipment, with an engineer reviewing and approving the work.

That's an important distinction. Industrial AI isn't simply about bringing powerful AI models into a factory. It's about combining AI with deep industrial context and engineering expertise so it can be deployed safely and reliably in the real world.

And Eigen is one part of a much larger effort. Siemens is investing $1 billion in AI capabilities and offerings over the next three years, building on decades of experience applying technology across industry and infrastructure.

Because the next chapter of AI won't be defined only by more powerful models. It will also be defined by our ability to deploy them in the real world—with the industrial context and human expertise required to solve real problems and deliver measurable value.

That's a quieter kind of progress than a chatbot getting cleverer. It happens inside factories, working alongside engineers and, ultimately, inside those PLCs most of us will never see.

But we'll all experience the results.

I'm Lauren Espin, and this is The Optimistic Outlook.