Chemicals and materials / Glass and chemical materials
AGC: Generative AI analysis of skilled workers' spoken work logs on the factory floor and creation of skill transfer materials
- Company
- AGC
- Country
- Japan
- Adoption stage
- Pilot
- Source published
- Date basis
- The date the source was published. It can differ from the date adoption started.
- How the source was checked
- Read the full source text
The work problem
According to a press release issued on October 15, 2026 by CEC, which co-developed the system, manufacturers urgently need to raise productivity and deal with labor shortages, and improvement through work analysis and the digitization of skilled techniques are drawing attention. CEC and AGC have jointly developed SmartLogger, which records work logs with smart devices, since 2016, but methods such as smart glasses mainly captured working time and movements. Tacit knowledge, such as what skilled workers look at and how they make decisions during work, was hard to record, leaving skill transfer as a challenge.
Technology and data
The two companies developed SmartLogger Voice, a speech recognition data collection system that gathers work records and on-site knowledge from what workers say while they work. Recording starts when a worker says a pre-registered keyword, and the system automatically collects only the information needed for analysis, such as worker information, work start and end times based on when audio was recorded, and work content, including what workers murmur about their observations and decision points. CEC handled the speech recognition function and the collection system, while AGC developed software that converts the collected voice data into text and generates data, and at an AGC manufacturing site the companies ran a trial in which generative AI analyzed and organized the data and automatically produced work analysis reports.
Results
According to the release, the trial cut the time needed to analyze skilled workers' tasks and to document skill transfer materials to one tenth or less of the previous level in each case. CEC said that turning skilled workers' tacit knowledge, which had been hard to put into words, into a form generative AI can readily use helps reduce the risk of losing knowledge and the burden of training successors.
Limits and open questions
The release does not disclose which plant hosted the trial, how long it ran, how many tasks were analyzed or whether the system has moved into regular operation.
Sources
- シーイーシーとAGC、生成AIを活用し製造現場の匠の技を次世代へ継承音声認識システム『SmartLogger Voice』を共同開発prtimes.jp, Accessed
Compiled from public sources. These are not results from ATF Works customers.