The missing layer: Why AI agents need more knowledge
Shane McArdle, Kongsberg Digital
In this compelling session, Shane McArdle, CEO of Kongsberg Digital, explored why many organisations are investing in AI agents yet still struggling to scale them into trusted operational systems. While adoption across the energy sector is accelerating, the gap between experimentation and real value creation continues to widen because organisations are overlooking a critical foundation: operational knowledge.
Drawing on Frederick Winslow Taylor’s work on scientific management, McArdle explained that industrial operations still depend heavily on undocumented human expertise and lived experience that rarely exist inside digital systems. In high stakes environments, operators do not trust recommendations they cannot verify, especially when safety and operational performance are involved.
The presentation introduced the idea of a missing “memory layer” within physical AI, focused on capturing domain knowledge, engineering truth, and the reality of how work actually gets done across industrial assets. Without this layer, AI systems may be intelligent, but they remain difficult to trust and scale effectively.
McArdle concluded by emphasising that successful AI adoption requires more than technology alone. Explainability, trusted data foundations, and human judgement built into workflows are essential for creating AI systems that organisations can confidently rely on in real operations.
