AI is rapidly moving from experimentation to execution. As organisations deploy copilots, agents and increasingly autonomous workflows, the conversation is shifting from what AI can do to what it costs to do it at scale. In this session, Steve Westgarth explores the emerging economics of enterprise AI and asks whether the cost of intelligence could become a constraint on the ambitions organisations have for it. Drawing on a real-world customer service case study from Haleon, he will examine how agentic workflows can deliver meaningful improvements while creating new challenges around token consumption, model choice, orchestration and the cost of increasingly complex AI interactions. The session will explore practical strategies technology leaders can use to understand and manage AI unit economics without constraining innovation, and consider what comes next as intelligence becomes a metered enterprise utility: consumed on demand, embedded everywhere and carrying a cost every time we use it.