Murata claims AI agent R&D platform lifts output 約22.5倍

Murata's materials development work generates a wide range of data — material composition, manufacturing conditions, analysis results and property evaluations. The article notes that such diverse data is sometimes managed in different formats by research theme or department, which can make searching and using accumulated in-house experimental data time-consuming. The new platform was built to address that, so that accumulated data can be used efficiently across research themes and organizations.

What the system does is connect existing assets — experimental and analysis databases, Murata's own algorithms and internal technical knowledge — with a set of AI agents. By linking these seamlessly to an analysis environment that uses AI and machine learning, Murata says engineers on different teams can readily use prior know-how on material composition and manufacturing conditions, and can more efficiently grasp trends in experimental results, analyze causes and search for optimal conditions. Multiple agents can be tied together by role.

The reported evidence is still limited in scope. Murata says an internal demonstration assuming materials development work confirmed the full sequence of tasks could be executed, with one case running from data loading through analysis and chart creation to report writing in about 10 minutes. The headline efficiency number, meanwhile, is a simulation against a set business model rather than a measured company-wide result. How much of that gain holds up outside materials development — where data formats and research themes differ — is likely to be the real test.