Manufacturing / Industrial control systems
Rockwell Automation: Generative AI troubleshooting assistant for equipment maintenance technicians at the Singapore plant (Maintenance Copilot)
- Company
- Rockwell Automation
- Country
- United States
- Adoption stage
- In operation
- Source published
- In use from
- 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 an article Microsoft published on September 24, 2026, technicians at Rockwell Automation's Singapore plant used to leaf through manuals hundreds of pages long to find an error code and its recommended fix when a machine stalled, and if that failed they had to track down a colleague who had solved a similar problem before. Plant director Li Wang said the company wanted to solve the loss of decades of tacit shop floor knowledge as workers retire, and to make troubleshooting consistent regardless of who is working on a problem.
Technology and data
Since October 2025, technicians overseeing hundreds of machines at the plant have used the GenAI-Powered Maintenance Copilot, an in-house AI assistant. It was built with Azure OpenAI in Microsoft Foundry on Microsoft Azure, runs on the GPT-5.4-mini model, and answers from a database of insights created by veteran engineers with 20 to 30 years of shop floor experience, data from manufacturing software and the machines' instruction manuals. When a machine error is detected, it shows the most likely causes organized by symptom, cause and reaction, and technicians ask questions in natural language on a tablet or through augmented reality goggles and receive step-by-step guidance.
Results
According to plant director Wang, this consistent, targeted approach has lowered machine downtime by 33 percent, and Rockwell's internal tracking shows servicing and spare parts costs down by about 25 percent. Bob Buttermore, Rockwell's chief supply chain officer, said that instead of taking nine months for new workers to learn how to troubleshoot hundreds of machines, it now takes three months. The company plans to roll out the copilot to its plant in Twinsburg, Ohio, as well as plants in Poland and Mexico.
Limits and open questions
The article does not give the period or baseline used to measure the reductions in downtime and costs.
Sources
- Rockwell Automation pairs AI with decades of shop floor know-how so workers can solve glitches fasternews.microsoft.com, Accessed
Compiled from public sources. These are not results from ATF Works customers.