This week's manufacturing news shows that AI agents are becoming the new normal in factories around the world.

What is happening with AI in manufacturing right now? Manufacturers are moving past just trying out AI in small test projects and starting to use it for real work. A new report shows that most big manufacturers now believe that smart manufacturing with AI will be their biggest way to stay competitive over the next three years. Companies are putting a lot of money into these projects—78% of manufacturers are now spending more than 20% of their improvement budget on AI and automation.

Why are companies struggling to make AI work? Even though companies want to use AI and have money to spend on it, they're having trouble actually getting it to work well. The main problem is that most factories have many different computer programs that don't work together automatically. For example, a factory might have one program for planning what to make, another program for managing machines, another for checking quality, and another for keeping track of supplies. When a machine breaks down or a supplier sends materials late, information doesn't flow automatically between these programs. Workers still have to copy information from one program and paste it into another by hand, which wastes time and causes mistakes.

What AI agents are working the best? The most helpful AI agents right now aren't the flashy ones that sound like humans—they're the ones that help with actual factory work. Production planning AI agents are helping factories schedule which products to make and when to make them. When something goes wrong, like a machine breaks, these AI agents quickly figure out a new schedule that still gets products to customers on time. According to research, factories using these agents are getting products to customers 30% more often on time, cutting planning work by 80-90%, and making 2-5% more money. Another helpful AI agent is predictive maintenance, which watches machines using sensors and tells workers when a machine is about to break so they can fix it before it actually stops working. General Electric's factory in Munich used this kind of AI on over 3,000 machines and could predict failures with 92% accuracy up to two weeks early, which cut surprise breakdowns by 25%. Quality control AI agents use cameras and smart computer vision to check every single product and find problems that human eyes might miss. One company using this method caught defects 99.7% of the time instead of the 85-90% that humans could do, which means way fewer products with problems reach customers.

A new AI tool for paperwork problems. One company announced this week that it created an AI data transformation layer to solve a huge problem in factories: the boring paperwork. In the United States, many factories still receive purchase orders, invoices, delivery notes, and shipping papers as PDFs, scanned images, or emails. Workers have to read these documents by hand, pull out the important information, check if everything is correct, and then type it all into the company's computer system. This takes a lot of time and causes mistakes. The new AI tool can read these messy documents and automatically turn them into clean information that goes straight into the factory's computer system. This saves weeks of work and helps factories pay their bills faster and have better information for making decisions.

Factories need to change more than just their computers. An expert working with factories pointed out that the biggest challenge isn't actually the AI technology—it's that most factories were built and designed before AI and lots of automation existed. When factories add a lot of new automated machines, the physical layout of the building becomes a problem. More robots and machines moving around means more traffic, which can cause problems if the building wasn't designed for all that activity. Companies like Samsung and Dassault Systèmes are working with AI companies to rebuild how their whole factories work, but many smaller factories are struggling because they don't have good quality data, don't have enough trained workers, or have computer systems that don't connect well. Experts say that the real advantage in 2026 won't go to companies that just have the best AI tools—it will go to companies that can use all their tools together across their whole business and get their workers excited about the changes.