Overview
A build-along workshop taking developers from an Azure OpenAI resource to a working AI application that answers questions grounded in private documents — not just a raw chat wrapper.
What we built
- Chat completions — calling Azure OpenAI from .NET with system prompts and streaming.
- Embeddings — turning documents into vectors and storing them for search.
- Retrieval — pulling the right context and grounding answers to cut hallucinations.
- Guardrails — content filtering, token budgeting, and graceful failure.
- Cost & scale — deployment tiers, quotas, and keeping the bill predictable.
Who attended
.NET developers shipping their first production AI feature. 140+ attended.