Skip to content

1.1 Learning Objectives

This is based on the Build a custom chat app with the Azure AI Foundry SDK tutorial but adapted for use in workshops with a devcontainer for fast setup. Fork the repo, launch GitHub Codespaces - and get started!

The lab teaches you to build a RAG-based copilot using the Azure AI Foundry platform By completing this lab, you will learn to do the following tasks:

  1. Use the Azure AI Foundry Portal to
    • start a new Azure AI project
    • discover and deploy AI models to the project
    • add an Azure AI Search resource to the project
    • setup an Application Insights resource for tracing
  2. Use the Azure AI Foundry SDK to
    • create a new search index with your data
    • extract customer intent with an intent mapping prompt template
    • retrieve relevant knowledge using the search index
    • create a chat agent with a grounded prompt template
    • process customer queries returning responses grounded in my data.
  3. Use the Azure AI Evaluation SDK to assess app quality by
    • creating a custom evaluation dataset.
    • executing a custom evaluation workflow.
    • viewing evaluation results, in your local environment.
    • viewing evaluation results, in Azure AI Foundry portal.

This sets you up with a "sandbox" project that you can use to explore other features or tooling in Azure AI Foundry. For example - try bringing your own application data, or adding secondary data sources for knowledge retrieval as part of the RAG flow.

BONUS: Use what you learn here to explore Contoso Chat next! (click to expand)

Contoso Chat is an open-source reference sample that implements the retail RAG-based copilot (using the same data) in a code-first approach from initial provisioning to final production deployment.

  • The current version (v3) provisions the Azure AI Foundry resources using the Azure Developer CLI (azd provision) and deploys the final application to Azure Container Apps for a hosted API endpoint.
  • The next version (v4) will incorporate more elements from the Azure AI Foundry SDKs that we highlight in this RAG tutorial, including use of the Azure AI Model Inference API and the Azure AI Evaluation SDK.