Panasonic Industry believes that using artificial intelligence in manufacturing does not begin with selecting a ready-made model or tool, but with building a reliable operational database. According to the article published in MONOist Japan, the company has spent more than ten years converting production-line data and the knowledge associated with it into a structure that can be used to improve operations.
This experience comes from a company founded in 1969 that produces, among other products, aluminum electrolytic capacitors. These components are used in areas including electric vehicles, artificial intelligence applications, and telecommunications infrastructure. In this context, the company says that improving production requires addressing differences among manufacturing lines, providing resources, and responding to changes in demand volume.
Data Came Before Artificial Intelligence
About ten years ago, nearly half of the older production equipment dating from before 2010 was unable to collect data directly. Panasonic Industry therefore worked to introduce equipment capable of supplying data and install programmable logic controllers, or PLCs, on existing production lines.
According to the article, this process made it possible to collect data from more than 600 production lines. The standardization process was not limited to general indicators; it became possible to organize data such as production quality, equipment defect rates, product defects, volume-related defects, production efficiency, and other operational indicators.
From Measurement to Operational Knowledge
The company does not view data collection as an end in itself. Making data available from all equipment helps analyze operating conditions and identify the causes of problems, while also enabling comparisons of equipment and process performance. Panasonic Industry connects data with the knowledge accumulated by employees, considering the practical question to be how to understand the relationship between what employees do and what changes in process performance.
The article indicates that the company is working to build an operational knowledge framework based on converting field experience into a reusable format, rather than leaving it tied to specific individuals. This point is particularly important in manufacturing environments where product specifications and production methods change from one customer to another and may require thousands of setup and operating procedures.
What Changes in Practice?
The actual transformation lies in moving from using artificial intelligence as a separate tool to integrating it with equipment data and operational knowledge. The article explains that Panasonic Industry has begun employing technologies including AI agents to support operations, but this step came after preparing the foundational data infrastructure and standardizing performance indicators.
The key takeaway is that artificial intelligence cannot compensate for missing data or inconsistent definitions. A factory that cannot access its equipment data, or does not have a unified way to describe defects and production quality, will face limitations before reaching the stage of selecting an artificial intelligence model. In Panasonic Industry’s case, digitizing more than 600 production lines provided a foundation for analyzing operations and expanding knowledge-based automation.
The article does not establish that all production problems have been solved or that AI agents have become a replacement for human expertise. Instead, it presents a clear path: collect data from equipment, standardize it, connect it with field knowledge, and then use artificial intelligence in manufacturing operations. The effectiveness of this path remains tied to data quality, the accuracy of knowledge modeling, and employees’ ability to verify the results before converting them into operational decisions.