Global Lighthouse Network 2026

3.2 Cognitive factories: staying value-backed with human-centric AI Human-in-the-loop decision making: balancing automation with oversightLighthouses are driving the shift from “smart” to “cognitive” factories – sites with self-learning, adaptive and increasingly autonomous operations. Powered by advanced AI applications such as AI agents, these factories enable a single operator to oversee multiple systems that are increasibly capable of making real-time decisions for core processes. Drawing on deep process expertise and rich data ecosystems, Lighthouses are building AI- enabled “brains” that orchestrate entire operations with speed and precision. AI applications demand robust safety and governance, but agentic AI – where systems act as active collaborators – requires even greater rigour. Lighthouses embed human oversight as a core design principle, moving beyond passive “human-in-the-loop” models. As agents begin managing other agents, it becomes critical to pinpoint moments for human intervention and define accountability. Fairness audits, explainability frameworks and continuous governance loops are now table stakes for maintaining trust as automation scales up, ensuring AI decisions are transparent and that leaders understand not just what the system decides, but why (Figure 22). Global Lighthouse Network: Rewiring Operations for Resilience and Impact at Scale 34