As AI adoption grows, costs and security risks can become buried across model choices, agent workflows and MCP tool connections. This session examines where unnecessary spend and over-provisioned access hide and how teams can gain the visibility needed to address them before they scale. You’ll learn how to: → Identify the models, agent behaviours and tool calls driving unnecessary spend. → Detect access risks created by overly broad MCP permissions. → Apply consistent controls across technical and non-technical teams. → Build a practical framework for managing AI cost, access and governance at scale.