Rather than moving your data into a separate search system, you'll learn how Azure SQL can combine relational data, full-text search, vector search, and AI models to deliver relevant, grounded results. You'll build the pipeline yourself, from embeddings and retrieval through ranking and generation. This matches the workshop's core goal of letting SQL developers own the end-to-end AI search pipeline.

Along the way, we'll cover:

• Preparing data for AI: Structure, chunk, and enrich existing relational data for retrieval.
• Embeddings and vector search: Generate embeddings, create vector indexes, and retrieve by meaning instead of exact words.
• Full-text and hybrid search: Combine semantic and keyword retrieval to overcome the limitations of either approach alone.
• Query rewriting: Turn natural-language requests into better searches, filters, and retrieval strategies.
• Ranking and reranking: Fuse results with techniques such as Reciprocal Rank Fusion and improve which results reach the model. The workshop specifically demonstrates why combining vector and full-text rankings produces more robust retrieval.
• RAG and agents: Use retrieved application data to ground AI responses and give applications and agents access to production-style data.

We'll use Azure SQL Database throughout, including its native vector type and indexes, full-text search, external models, and integration with Microsoft Foundry.

By the end of the workshop, you'll have:

• A working end-to-end RAG retrieval pipeline
• Practical experience with vector, full-text, and hybrid search
• A clear understanding of embeddings, query rewriting, ranking, and RRF
• An AI application grounded in real relational data
• Reusable patterns for building reliable, scalable AI applications and agents

Is this workshop for you?

Yes, if you know SQL and want to move beyond AI demos into applications grounded in real data.

You don't need to be an AI expert. If you understand basic SQL and have some familiarity with prompts, embeddings, or RAG, you're ready.