Build a RAG-based Retail Copilot Code-First on Azure AI Foundry
summary
Code-first recipe that deploys a running retail copilot from an open-source sample: Contoso Chat uses Azure AI Foundry, Azure AI Search for 20 product items (8 categories) and Azure Cosmos DB with 12 customer profiles, deploys on container apps, and wires embeddings + chat completions (GPT‑4o‑mini and text‑embedding‑ada‑002 in the demo). Iteration happens via Prompty (editable prompt templates, test runtime, convert-to-Python), provisioning via ACD up and ACD deploy, includes safety guardrails, an evaluation notebook that grades groundedness, and a path to enable vector search in Cosmos DB.
AI-written summary · about
chapters (9)
- 0:00Introduction to Contoso chat
- 1:11Scenario and data analysis
- 3:50Architecture and RAG design
- 5:40Development lifecycle overview
- 7:25Quickstart and environment setup
- 11:55Ideation with Prompty
- 21:03Evaluation and testing
- 23:07Deployment and operationalization
- 24:10Sandbox exploration and conclusion
related talks
01h 58mBuild, Evaluate and Deploy a RAG-Based Retail Copilot with Azure AI
45mBuilding a copilot: Azure AI Studio or Copilot Studio
46mBuild and maintain your company Copilot with Azure ML and GPT-4
45mThe era of the AI Copilot
56mBuild your own CoPilot
58mNext generation AI for developers with the Cloud
36mWhat's new with Copilot Studio
42mCode First LLMOps With Azure AI & Prompty