AI Assistants join the factory layer

Main machine It has been the same since 1900 for a steel ball carrier, but the manufacturers automated everything around in a steady way. Today, the process is governed by the conveyor belt and is automatic for most. The most relevant task for people is to understand that things go wrong and even handed over to AI.

The Schaeffler factory in Hamburg starts with a steel wire, cut and rude balls. These balls are tougher in a number of stoves and then put three more accurate grinder until a part of the microns is as spherical. The result is one of the most versatile components of the modern industry, provides low friction joints from tbers to car engines.

The accuracy level requires a permanent test, but when the flaws are corrected, you can present a puzzle that follows them. The test can display a defect in one time in the mounting line, but the reason may not be known. Perhaps the torque on a screw instrument or a newly modified grinding wheel affects quality. It means to follow the problem, to compare data between many industrial equipment, none of them are taken into account.

It can also be a business for cars soon. Last year, Schaeffler was one of the first users of Microsoft factory operations agent, was equipped with large language models and was one of a new product designed specifically for producers. ChatBot style means to monitor the causes of defects, work hours or excess energy consumption. The result is something like ChatGPT for factories, Openai is used in the background thanks to the company’s partnership with Azure of Microsoft.

Microsoft vice-president of Global Industrial Marketing Kathleen Mitford describes the project as a “an agent who gives a basis of production information at the beginning of production data.” As a result, Mitford says: “Agent is able to understand the questions and translate accuracy and precision against standard data models.” Thus, a factory employee can ask a question like “causing things higher than the usual defects”? And the model can respond to data throughout the production process.

Agent, Microsoft’s existing enterprise products, especially Microsoft Fabric, deeply united into the information analytical system. This means that Hundreds of plants operating in the Microsoft system are able to inform the agent from all over the world.

Stefan Soutschek said that Schaeffler is the vice president of the defendant, the system of data analysis system. “Although the main benefit, it is not the conversation itself,” he says. “This is a combination of this grass (operating technology) data platform and trust in this information.”

Despite its name, this agent is not AI: there are no goals and the powers are limited by answering the user’s questions. Microsoft can set the agent to perform the main commands through Copilot Studio, but the goal is not the lack of agent who decides. This is primarily as a means of obtaining information AI.