Overview

This exercise demonstrates how to create an AI agent using Azure AI Foundry. The agent assists employees with expense claims by leveraging a GPT-4o model and grounding responses in a corporate expenses policy document.

Steps & Configuration Details

1. Create an Azure AI Foundry Project

  • Open Azure AI Foundry portal (https://ai.azure.com) and sign in.
  • Select Create an agent.
  • Configuration Items:
    • Azure AI Foundry Resource: A valid name.
    • Subscription: Your Azure subscription.
    • Resource Group: Select or create a resource group.
    • Region: Choose any AI Services-supported location.
    • Project Name: A unique name.

After creation, the Agents playground opens automatically, with a GPT-4o base model deployed.

2. Configure the AI Agent

  • Set the Agent Name to ExpensesAgent.
  • Ensure the GPT-4o model deployment is selected.
  • Define Instructions:
    You are an AI assistant for corporate expenses. You answer questions about expenses based on the expenses policy data.
If a user wants to submit an expense claim, you get their email address, a description of the claim, and the amount to be claimed and write the claim details to a text file that the user can download.

3. Add Knowledge to the Agent

  • Download the expenses policy document:
    https://raw.githubusercontent.com/MicrosoftLearning/mslearn-ai-agents/main/Labfiles/01-agent-fundamentals/Expenses_Policy.docx
  • Upload the document to Azure AI Foundry:
    • Select + Add Knowledge → Files.
    • Create a Vector Store named Expenses_Vector_Store.
    • Upload and save Expenses_Policy.docx.

Configuration Items:

  • Vector Store Name: Expenses_Vector_Store
  • File Count: 1

4. Add Actions to the Agent

  • Select + Add Action → Code Interpreter.
  • Save the configuration.

The Code Interpreter allows the agent to generate and execute Python code for processing expense claims.

5. Test the AI Agent

  • Open the Playground Chat.
  • Submit the prompt:
    What's the maximum I can claim for meals?
    • The response should be based on the uploaded expenses policy document.
  • Submit a follow-up prompt:
    I'd like to submit a claim for a meal.
    • The agent should request an email address.
  • Provide an email (e.g., [email protected]).
  • Submit a claim description and amount:
    Breakfast cost me $20.
    • The agent should generate a text file with claim details and provide a download link.

6. Clean Up

  • Delete Azure resources to avoid unnecessary costs:

Related Posts