This session introduces the emerging concept of Agentic Business Intelligence, where semi-autonomous software agents collaborate with human stakeholders to design, build, and evolve BI systems. Leveraging agent memory, reasoning, and goal-orientation, these systems transform traditional metadata-driven automation into dynamic, adaptable workflows. Using frameworks like SPAR (Sense, Plan, Act, Reflect) and agent frameworks such as n8n, Autogen and others, we will discuss how agents can help maintain glossaries, models, and pipelines while enabling teams to continuously align data delivery with business outcomes. This process bridges the conceptual and physical layers of BI (using metadata) to show how agentic approaches are not just futuristic—but "deployable" today.