Basic Machine Bearings to polish steel ball bearings have been the same since around 1900, but manufacturers have been steadily automating everything around. Today, the process is driven by a conveyor belt and is in most cases automatic. The most urgent task for humans is to figure out when problems arise – it can be handed over to AI.
The Schaeffler factory in Hamburg starts with steel wire, cut into rough balls. The balls harden in a series of furnaces and then pass through three increasingly precise grinding stones until the ball reaches one tenth of the micron. The result is one of the broad components of modern industry, from lathes to automotive engines, which enable low friction joints.
Accuracy requires continuous testing, but tracking them can be difficult when defects do occur. The test may show that a defect occurred at some point on the assembly line, but the reason may not be obvious. Maybe the torque on the spiral tool has been turned off, or the newly replaced grinding wheel will affect the quality. Tracking the problem means comparing data across multiple industrial equipment, none of which takes this into account.
For machines, this may soon become a job. Last year, Schaeffler became one of the first users of Microsoft’s factory operations agents, a product powered by large language models designed specifically for manufacturers. Chatbot-style tools can help track the causes of defects, downtime, or excess energy consumption. The result is similar to the factory Chatgpt, which thanks to the company’s partnership with Microsoft’s Azure, Openai’s models are used on the backend.
Kathleen Mitford, vice president of Microsoft’s global industry marketing, described the project as “a reasoning agent that operates on top of manufacturing data.” As a result, Mitford said: “Agents are able to understand the problem and translate standardized data models with accuracy and accuracy.” So factory workers might ask a question like “what causes higher defects than usual.” What?” The model will be able to answer through data throughout the manufacturing process.
The agent is deeply integrated into Microsoft’s existing enterprise products, especially Microsoft Fabric, its data analytics system. This means that Schaeffler runs hundreds of factories on Microsoft’s systems and is able to train its agents on data around the world.
Stefan Soutschek, vice president of IT at Schaeffler, said the scope of data analytics is the real power of the system. “The main benefit is not the chatbot itself, although it helps,” he said. “This is a combination of this Old Testament. [operational technology] The data platform and chatbots on the backend depend on this data. ”
Despite its name, this is not a proxy AI: it has no goals and its capabilities are limited to answering any questions asked by users. You can set up a proxy to execute basic commands through Microsoft’s Copilot Studio, but the goal is not to let the proxy make its own decisions. This is mainly AI as a data access tool.
