The company reports pilots that adjusted equipment settings, while keeping operating boundaries and human responsibility central.

Industrial AI can influence physical equipment, so AVEVA says automation needs tighter controls than software used only on a screen. Arti Garg, the company’s chief technologist, described responsible AI as secure, efficient, and focused on human safety and oversight.

Garg said AVEVA wants AI to “augment” rather than replace people in critical decision loops. One system takes operational data and simulations of industrial processes, then recommends a new set point—the target operating level for equipment—as plant conditions change.

AVEVA also reported successful real-world pilots that automatically adjusted industrial set points. Garg said guardrails could keep those changes within a defined operating band or limit automation to specific parts of a facility. The discussion does not explain how often people approved or overrode automatic changes.

Details that would help assess those pilots—including sites, durations, comparison baselines, and safety outcomes—were not disclosed in the supplied discussion. That makes the results useful as company-reported examples, but not a complete basis for judging wider deployment.

Garg said AVEVA is still “trying to figure out” how human operators should become supervisors of increasingly automated systems. The practical model today is bounded automation: AI can act inside defined limits while people retain responsibility for the larger decision.

AVEVA also described work with Idaho National Laboratory to help grid operators identify unusual behavior and respond earlier. The source presents this as ongoing project work, not a confirmed large-scale deployment or measured result.

For SCG Chemicals in Thailand, AVEVA says its tools combine operational and engineering data with proprietary models that detect anomalies. Garg reported a 99% reliability target and almost ninefold return on investment during the platform’s early pilot period. Those figures came from an AVEVA-sponsored MIT Technology Review Insights discussion, so site-specific validation remains the next step before expanding automation.