Your organization wants "AI on our data" — but shipping your data to yet another specialized vector database is a hard sell. SQL Server 2025 changes the conversation: vectors are now a native type, DiskANN gives you approximate nearest neighbor search at scale, and functions like AI_GENERATE_EMBEDDINGS and AI_GENERATE_CHUNKS handle the plumbing, all from T-SQL.

In this session we'll build a real semantic search and RAG scenario on business data, step by step: chunking documents, generating and storing embeddings next to the relational data they describe, creating vector indexes, and combining vector similarity with plain old WHERE clauses — the thing dedicated vector stores struggle with. Then we'll wire it into an AI agent from VS Code and see the full loop working.

We'll close with the architect's view: when database-native vector search is the right call, when a dedicated store still wins, and how mirroring to Microsoft Fabric fits into the picture for analytics on the same data.

Extended version available (up to 120 min) with a live end-to-end agent demo.

Level: Advanced (300). Duration: 60–90 min (extended 120-min version with live agent demo available). Session can be delivered in English, Spanish or Portuguese.