JapanLogistics and transport2025 – ongoingEast Japan Railway Company (JR East)
Generative AI recovery support for signal and telecom equipment faults
JR East put generative AI into its fault-recovery system for signal and telecom equipment and expects to roughly halve recovery time
Problem
When signal or telecom equipment fails, dispatchers at the control center must work out the cause and tell field staff what to do. Complex faults take time, and the control center carries a heavy load.
Approach
JR East built the system with BIPROGY and announced it on June 10, 2025. A generative AI model learned from past fault timelines and manuals. It logs the radio talk between field staff and dispatchers in time order and shows a likely cause, a response plan and an estimated recovery time. An earlier machine-learning version had been in partial use since March 2023. A pilot on the Tokyo-area transport management system was planned for September 2025, and a radio speech-to-text summary function for fiscal 2025.
Results
Takeaway
BIPROGY said the generative AI version needed less training time and cost less than the older machine-learning one. The 50% cut in recovery time remains a company expectation.
Categories
- Size
- Large enterprise
- Industry
- Logistics and transport
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Generative AI
- Adoption
- Co-developed with a partner
- Evidence
- Company's own claim
Sources
Source summaries
Describes a generative AI system that transcribes radio calls during signal and telecom faults and proposes a cause, response and recovery time. It says a pilot on the Tokyo transport management system was due in September 2025 and that an earlier machine-learning version ran from March 2023. The 50% shorter recovery time is given as an expectation.
Reports BIPROGY's announcement that it delivered a generative AI recovery support system to JR East. The AI learned from accumulated fault timelines and manuals, and BIPROGY said it trains faster and costs less than machine-learning AI. A radio speech-to-text summary function is due within fiscal 2025, and JR East plans a rail-specific generative AI by the end of fiscal 2027.
Full details are in the original sources.
