Living Digital Twins Bring Agentic AI and Human Control to Factories
Living digital twins and agentic AI promise safer, leaner factories, but AVEVA says human oversight and stronger governance remain essential in industry.
Summary
Living digital twins merge dynamic plant models, engineering and design records, live IoT and control telemetry, and predictive and physics based models. Unlike earlier simulation tools, they can detect anomalies and intervene through operations and control systems. Operators can assess emissions, asset health and output. AVEVA chief product officer Rob McGreevy says agentic AI enables bounded action, while engineers define healthy operation, monitor execution and catch errors. He expects SCADA style human supervision, with full autonomy limited to narrow, repetitive tasks. Workers shift from single tasks and routine approvals to multidisciplinary limit setting.
A 2026 Manufacturing Leadership Council survey found 66% of manufacturers use or plan agentic AI and 40% expect some factory autonomy by 2030. Deloitte’s 2025 survey found 65% of executives rank operational risk among their top concerns. AVEVA and IMD found AI changed partner accountability significantly or fundamentally at 23% of organizations, slightly or not at all at 43%, while only 27% shared data substantially or extensively. The EU AI Act treats certain critical infrastructure safety AI as high risk and requires human oversight. AVEVA has a responsible AI board and published principles.
Saudi Aramco’s Yanbu refinery uses supervised closed loop optimization, while Ras Tanura operators approve algorithmic recommendations. Nestlé cut Nesquik dryer energy use 16% while improving quality and yield. SCG Chemicals’ engineering, operations and maintenance twin reports 99% reliability and ninefold ROI within six months. The National Laboratory of the Rockies uses generative AI for renewable forecasting and grid condition estimates. McGreevy says drones, robotics and physical AI could inspect pipelines and power lines for hazards, reducing exposure, incidents and outages. After 50 years of industrial silos, twins could coordinate value chains around supply, demand, tariffs and material availability.
Positives
- 66% of manufacturers use or plan to use agentic AI, showing strong demand for more autonomous industrial operations.
- Nestlé reduced Nesquik dryer energy consumption by 16% while improving product quality and yield.
- SCG Chemicals reports 99% plant reliability and a ninefold return on investment within six months.
- Saudi Aramco’s Yanbu refinery already uses optimization algorithms to adjust production in a supervised closed loop.
- Physical AI and drones could inspect hazardous pipelines and power lines while reducing human exposure, incidents and outages.
Risks & concerns
- 65% of executives ranked operational risk, including possible business disruption, among their leading smart manufacturing concerns.
- 43% of organizations said AI changed accountability among ecosystem partners only slightly or not at all.
- Only 27% of organizations reported substantial or extensive data sharing with partners, limiting coordinated industrial intelligence.
- Full AI autonomy still requires more testing, validation and proof points, restricting near term deployment to narrow, repetitive tasks.
- AI can reinforce patterns in flawed data, making early human detection of misaligned actions essential.