Automation

Rockwell Singapore plant cuts defects 35% with AI maintenance

World Economic Forum Lighthouse facility uses machine learning, generative AI copilots, and multi-agent systems to transform operations.

Omega Editorial · November 19, 2026 · 3 min read

Key takeaways

  • Rockwell Automation's Singapore facility reduced defects 35% using a multi-agent AI quality system that integrates machine, inspection, and metrology data.
  • A generative AI maintenance copilot cut technician training time from 12 months to four months and reduced mean time-to-repair from 18 to 12 minutes.
  • Predictive maintenance using 64 machine-learning algorithms has eliminated conveyor-related production disruptions over two years of operation.
  • The facility achieved 25% reduction in spare parts inventory and over $500,000 in transportation savings through AI optimization.
  • The World Economic Forum designated the plant a Global Lighthouse in June 2026 for its use of Fourth Industrial Revolution technologies.

Rockwell Automation's Singapore manufacturing facility has achieved a 35% reduction in defects and cut technician training time by two-thirds using artificial intelligence systems across maintenance, quality control, and workforce development, according to updates shared at Automation Fair 2026.

The facility, designated a World Economic Forum Global Lighthouse in June 2026, operates as a "Rockwell on Rockwell" proving ground where the company tests technologies in its own operations before deploying them to customer sites. The Singapore plant was selected for the Global Lighthouse Network, an initiative that recognizes industrial facilities using Fourth Industrial Revolution technologies to advance productivity, resilience, and sustainability.

Predictive maintenance eliminates conveyor downtime

The facility aggregates hundreds of thousands of data points from controllers and its manufacturing execution system to feed 64 machine-learning algorithms monitoring a critical conveyor system that runs between two levels of the space-constrained plant. Since implementing the predictive-maintenance system two years ago, the conveyor has not caused a single production disruption, Cyril Perducat, senior vice president and chief technical officer at Rockwell Automation, reported during the conference.

A generative AI maintenance copilot combines plant data, MES information, equipment manuals, and knowledge captured from experienced technicians. Workers query the system using text or voice. The system reduced technician time-to-competency from approximately 12 months to four months and cut mean time-to-repair from 18 minutes to 12 minutes. The facility has also achieved a 25% reduction in spare parts inventory.

Multi-agent quality system drives defect reduction

The Singapore site deployed a multi-agent quality-assurance system that integrates data from machines, control systems, visual inspections, vibration analysis, metrology, and other sources to identify issues and recommend actions to technicians. This system delivered the 35% defect reduction since implementation.

Beyond the factory floor, Rockwell Automation applied AI to analyze employee transportation data across the region, optimizing bus routes for facilities including Singapore and generating more than $500,000 in transportation savings.

Why it matters

The Singapore facility demonstrates how manufacturers can layer multiple AI approaches—machine learning for prediction, generative AI for knowledge access, and multi-agent systems for decision support—to address distinct operational challenges. The results show measurable improvements in equipment reliability, workforce development speed, and quality outcomes, providing a reference architecture for industrial AI deployment that extends beyond pilot projects to sustained operational gains.

Perducat emphasized that successful automation requires understanding where users are within each process and using that knowledge to inform next steps, treating autonomation as a continuous journey rather than a fixed implementation. He described these technology-enabled efforts as "uniquely human" work that connects seemingly unrelated factors based on experience to develop new operational approaches.

"Either you resist change or embrace change. And we want to embrace change with confidence," Perducat said. "We are building what the future of factory operations looks like."

These details were first reported by Control Global in coverage of Automation Fair 2026.

  • #predictive maintenance
  • #generative ai
  • #manufacturing ai
  • #rockwell automation
  • #quality control
  • #industrial automation

This is an original analysis by the Omega editorial team. Source reporting: Automation Watch.

Want systems like this working for your business?

Book a Call

More in Automation