Design Enterprise-Grade RAG Systems with LLMs, Vector Search

Most agentic AI projects look impressive in a demo and then stall in production — not because the model is weak, but because there is no shared, governed understanding of what “customer,” “order,” or “incident” actually means across systems, and no way to trace an answer back to where it came from. This workshop closes that gap by treating the enterprise graph, not the prompt, as the foundation of a trustworthy AI agent. Over four hands-on hours, you will design and build a trusted AI/data agent: an ontology and graph modeled in Microsoft Fabric IQ, bootstrapped from your existing Power BI semantic models and grounded in real-time operational data, that a chat agent queries to give explainable, traceable, governed answers. Fabric IQ, Fabric Graph, Real-Time Intelligence, Purview, and Power BI are the tools you use to build the agent — the agent itself, and the trust it earns from business users, is the point of the workshop. How this workshop will accelerate your career By participating in this masterclass, you'll learn how to: • Design a trusted AI/data agent architecture grounded in a governed enterprise graph, not a vector index alone. • Model an enterprise ontology and knowledge graph in Microsoft Fabric IQ, bootstrapped from Power BI semantic models and enriched with real-time operational data. • Ground a conversational agent in that graph so every answer is explainable and traceable back to a specific entity or relationship. • Evaluate trade-offs between graph-grounded retrieval and classic vector RAG, and combine both where each is strongest. • Apply governance, guardrails, and reliability practices — Microsoft Purview, row- and column-level security, Responsible AI checks, human-in-the-loop approval — to move an agent from prototype to production. • Reduce hallucinations and test failure modes using a graph's explicit structure as a built-in evaluation and audit trail. Why attend now? Most GenAI and agentic AI projects fail not because of the model itself, but because of weak grounding, no shared business vocabulary, and a lack of governed, production-ready architecture. As enterprises move from chatbots to agents that reason over live enterprise data and take action, the skill in short supply is not prompt engineering — it is knowing how to model the graph an agent stands on, and how to govern what it is allowed to see and do. Microsoft Fabric IQ — announced at Microsoft Ignite in November 2025 and rolling out through 2026, with core agent capabilities reaching general availability in mid-2026 — is Microsoft's answer to exactly this problem: a shared context layer of business ontologies and graphs, sitting on OneLake, that both people and AI agents reason against. Getting hands-on experience with it now, while the capability is still forming, is a genuine early-mover advantage for AI engineering, data, analytics, architecture, and enterprise technology roles. Who should attend? This workshop is ideal for: • AI Engineers and GenAI practitioners • Data Engineers and analytics professionals • Software Engineers and solution architects • Data Scientists and Machine Learning Engineers • Technology Consultants and enterprise architects • Professionals transitioning into enterprise AI and GenAI roles Prerequisites • Basic understanding of Generative AI, LLMs, or enterprise data workflows • Familiarity with APIs, databases, or cloud-based data platforms is helpful but not required • Experience in data, analytics, AI, software engineering, or solution architecture will help you maximize your learning • No advanced machine learning or deep learning background is required • An interest in building practical RAG applications, grounded AI systems, or enterprise GenAI solutions Learn from the best Brian Bønk is a Data Platform MVP and Microsoft Recognized FastTrack Solution Architect with more than two decades of experience in data, analytics, and enterprise platform implementation. Known for bridging business and technology, he helps teams design scalable data solutions that deliver measurable impact. In this masterclass, Brian brings practical expertise across Microsoft Fabric, Power BI, governance, and AI architecture to help learners build grounded, production-ready GenAI systems with confidence and clarity in practice today.