Digital twins have circulated through manufacturing for years. Many started as photorealistic factory replicas that impressed in a demo and then sat unused. The real payoff appears only when a twin is scoped to one production line, one recipe change, or one safety question instead of an entire facility.

Tony Trim, Manufacturing Operations AI Leader at Eaton, opens the second installment of a three-part series with Stephen Hooper, Senior Vice President of Operation Solutions at Autodesk, describing a shift from digital twins as showpieces to twins built around a single decision. Trim ties that discipline to a roughly $50 million annual savings target across Eaton’s facilities. A paint line model that saved a couple hundred thousand dollars a year at one plant became an early proof point.

The conversation then moves from where twins create value to how AI builds on that foundation. Trim lays out four tiers of manufacturing intelligence: describe, diagnose, predict, and prescribe. Trim argues agentic AI finally lets organizations act on that fourth tier instead of stopping at analysis.

• Scope twins to one decision: Trim ties measurable value to models built around a single line, process, or recipe rather than a whole factory.
• Prioritize capital avoidance: Trim ranks getting more from existing assets above predictive maintenance, which needs failure history to mature.
• Engineer boundaries before agents act: Trim links agentic AI’s success on the plant floor to interlocks, clear human authority, and bounded, supervised actions.

Did you miss the previous episode?

On the move? Take this conversation with you. If you don’t have time to finish the video, you can stream the complete, uninterrupted audio edition of this episode on the go.