The Factory That Heals Itself
Living digital twins are giving industrial plants something close to a nervous system — with humans still holding the pen.
By The energy-climate Desk
A refinery, it turns out, has a metabolism. It ages, develops irregularities, and signals distress in ways that are easy to miss — until now. A new class of AI-infused digital twins, built to mirror physical reality in real time, is giving operators something earlier systems never could: the ability to intervene, not just observe. [1]
Where older digital twins focused on simulation, observation and what-if scenario planning, living digital twins combine a dynamic, high-fidelity digital representation of complex operations with the capacity to intervene in the physical world through operations and control systems. [1] Agentic AI pushes that further still — enabling industrial systems to move beyond diagnostics or suggestions and act of their own accord, as the technology progresses toward a kind of industrial homeostasis, in which small, constant adjustments keep a plant or factory healthy and humming. [1]
The commercial appetite is real. A 2026 Manufacturing Leadership Council survey found 66% of manufacturers are using or plan to use agentic AI in their operations, and 40% expect some degree of factory autonomy by 2030. [1] Yet in a Deloitte 2025 smart manufacturing survey, 65% of executives ranked operational risk among their top concerns. [1] Meanwhile, an AVEVA and IMD study found 43% of organizations said AI has changed decision-making and accountability across ecosystem partners only slightly or not at all. [1]
On the ground, the spectrum of trust varies widely. At Saudi Aramco's Yanbu refinery, optimization algorithms adjust production in a closed loop, with operators overseeing as needed. At its Ras Tanura refinery, the algorithms recommend changes and operators decide whether to implement them. [1] At Nestlé's Nesquik plants, AI tuning of industrial dryers reduced energy use by 16% in a landmark test, while improving quality and yield. [1]
The question was never whether machines could act faster than humans. It's whether the two can build enough mutual legibility to act wisely together — and what governance structures make that possible.
McGreevy sees AI working mostly in an advisory mode for the foreseeable future, with full autonomy limited to narrow, highly repetitive tasks — and, as he puts it, 'the "S" in SCADA means supervisory, and supervisory means there's a human in the middle.' [1] In a world of turbulent supply chains and shifting regulation, that balance may be less a constraint than the point.