AI Assistants Join the Factory Floor

Basic machine Steel ball bearing was similar to about 1900, but the manufacturers have been continuously automatically automatically. Today, the process is powered by a conveyor belt and mostly it is automatic. The most urgent task for people is to determine when things are wrong – even it can be handed over soon AIIThe

Hamburg’s Shaffler factory begins with steel cable that is cut and pressed on rough balls. These balls are tightened in a series of furnaces, and then the three growing specific grinders are left until they are rounded in the tenth of a micron. The result is one of the multipurpose components of modern industry, which enables short-lived joints in everything from lathe to car engine.

The accuracy of this level needs to be constant testing – but when the flaws are turned on, they can present a puzzle that track them. An error may occur at any stage of the Testing Assembly line, but the reason may not be obvious. Perhaps the torque of a screwing equipment is off, or a new replacement is affecting the quality of the grinding wheel. Looking for the problem means comparing data across multiple pieces of industrial equipment, which is not designed to keep it in mind.

It can be a job for machines soon. Last year, Shaffler became the first user of the factory operation agent in Shaffler Microsoft, it was a new product powered by big language models and specially designed for manufacturers. Chatbot-style equipment can help detect errors, downtime or additional energy usage. The result is something like the ChatzPT for the factory, the Openai models were used in the back -end to the company’s partnership with Microsoft’s Azur.

Global Industry Marketing Microsoft Corporate Vice President Cathlene Mitford describes the project as “a rational agent that works at the top of production data.” As a result, Mitford says, “Agent is able to understand the questions and translate them with accuracy and accuracy against quality data models.” So a factory worker “What is the reason for the general level of error?” Can you ask a question like this? And the model will be able to answer with data throughout the production process.

The agent has been deeply integrated into the existing enterprise products of Microsoft, especially on the Microsoft Fabric, its data analytics system. This means that hundreds of plants, which are capable of training his agent from all over the world, are on the microsoft system.

Stepan Soutskk, vice president of Shaffler, who is in charge of the IT, says the opportunity to analyze data is the real power of the system. “The main advantage itself is not a chatbot, though it helps,” he said. “It’s a combination of this OT [operational technology] The data platform and chattobot on the back -end depends on that data “

Despite the name, this is not agent AI: it does not have a goal and its strengths are limited to answering what users ask. You can set up the agent to execute basic commands through the Copilot Studios of Microsoft, but the target should not be decided by the agent’s own. This is basically AI as data access equipment.