Manufacturing / Industrial control systems
Hitachi: Searching past trouble cases and drafting response reports in the quality assurance department
- Company
- Hitachi
- Country
- Japan
- Adoption stage
- In operation
- 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 Hitachi published on June 4, 2026, global labor shortages, generational turnover and increasingly advanced and complex products and systems have made it more important for quality assurance departments in manufacturing to respond quickly and accurately when equipment failures or quality troubles occur. Yet trouble response and root cause analysis often depended on the experience and judgment that skilled staff had built up over many years. Hitachi's Omika Works, which develops control systems for social infrastructure, has many systems that run for more than 10 years and has accumulated large volumes of maintenance and operations information on customer questions and troubles in its databases, but responding to troubles and inquiries depended on the skilled staff who knew this information.
Technology and data
Hitachi's AI professionals and skilled frontline staff jointly developed an AI agent called the Quality Knowledge System that users can converse with in natural language. They analyzed which keywords and sequence skilled staff used to search for information and make judgments, built more than 100 items of business knowledge into the AI, and repeatedly refined the prompts and tuning by applying the data management know-how of Hitachi's in-house Generative AI Center. Staff search past trouble response records and manuals with natural-language questions or the text of customer inquiry emails, and create draft response reports by entering keywords such as the event, the phenomenon and the countermeasure policy. Hitachi first deployed the agent at Omika Works in Hitachi City, Ibaraki Prefecture, as the first user verifying its effectiveness itself, and has offered it to manufacturers as part of the HMAX Industry lineup since April 2026.
Results
According to the press release, at Omika Works the time to search for cases when a trouble occurs fell from 60 minutes to 5 minutes per case, a reduction of about 90%. The time to write a response report fell from 120 minutes to 15 minutes, and the time to analyze the root cause of a defect fell from 16 hours to 3 hours, each a reduction of more than 80%. Hitachi explained that even less experienced staff can draw on the know-how of skilled staff to make a quick initial response, and that reducing the dependence of report writing on particular people and the variation in its quality makes it possible to report to customers sooner.
Limits and open questions
The press release does not disclose the month adoption started at Omika Works, the measurement period and number of cases, or the AI model used. A function to analyze multiple quality documents is to be added in the future.
Sources
- 日立、熟練ノウハウを形式知化し、品質保証業務を大幅に効率化するAIエージェントをHMAX Industryとして提供開始prtimes.jp, Accessed
Compiled from public sources. These are not results from ATF Works customers.