Using Microsoft Fabric to orchestrate specialised AI agents across analytics, operations and business workflows

The session shows how organisations can move beyond single purpose copilots to teams of specialised AI agents grounded in trusted enterprise data, governance and business context. It includes the latest announcements and about half the time is live demos.
    Agenda (45 min)
Why single agents stall: the limits of one-assistant designs, and the anatomy of an "AI squad"
The foundation: grounding, orchestration, shared context and governance; Microsoft Fabric (OneLake, semantic models, real-time data) as the shared intelligence layer; what's new
Demo 1 - Build the Analyst: create and ground a Fabric data agent, inspect its reasoning and evaluate it against ground truth
Demo 2 - The squad in action: an orchestrator coordinates specialist agents to answer one business question, and two users get answers scoped to their own access
Demo 3 - From answering to acting: an operations agent detects an anomaly on live data, investigates the root cause and recommends an action for human approval in Microsoft Teams
Trust, takeaways and Q&A: identity, policy, observability and practical architecture patterns

    Audience: AI and data leaders, enterprise architects, data engineers and AI developers (Level 200–300). No prior Fabric experience is needed.