Build, Evaluate and Deploy a RAG-Based Retail Copilot with Azure AI
summary
Backend walkthrough that builds a production RAG retail copilot using Azure AI Studio and prompt flow, with code run in GitHub Codespaces and data in Cosmos DB + Azure AI Search. Workshop provisions 11 Azure resources, indexes ~12 sample customers and product markdowns, publishes the prompt flow as a container endpoint, and wires it to GPT-3.5 (with GPT-4 used for evaluation). Evaluation pipeline asks GPT-4 to score groundedness and catches hallucinations (toothbrush example), showing the deployed flow actually works end-to-end.
AI-written summary · about
chapters (10)
- 0:00Building a RAG retail copilot
- 10:59RAG workflow architecture
- 22:45Development environment setup
- 36:28Azure resources and data setup
- 56:03Prompt flow components
- 1:03:50Debugging prompt flow locally
- 1:17:09Deploying prompt flow endpoint
- 1:31:02Evaluating RAG performance
- 1:41:42LLMOps lifecycle improvements
- 1:57:03Q&A
related talks
27mBuild a RAG-based Retail Copilot Code-First on Azure AI Foundry
46mBuild and maintain your company Copilot with Azure ML and GPT-4
45mBuilding a copilot: Azure AI Studio or Copilot Studio
43mBuild and Maintain your Company Copilot with Azure ML and GPT-4 Q&A
58mNext generation AI for developers with the Cloud
40mAI Studio - Creating and scaling your custom copilots
48mGetting started with generative AI using Azure OpenAI Service
01h 01mUnlocking the Potential of AI