Two Problems Standing Between You and Industrial AI at Scale

Manufacturing and industrial operations are stuck in AI pilot mode despite years of Industry 4.0 promises. Two core challenges block progress: first, the lack of an integrated digital twin — a unified, contextualized data layer across all OT systems — without which AI can only optimize in silos. Second, and harder, is the inability to fully model the physical twin from digital data alone; experienced operator knowledge and physical-world context cannot be inferred from historical data. The authors outline three stages of AI autonomy (assistance, collaboration, autonomy) and argue that most industrial operations are still at stage one. Bridging the IT/OT divide — analogous to but far deeper than the Dev/Ops gap — is a prerequisite for any meaningful AI scale-up in manufacturing.