Skip to content

01 Exercise 🛠️: Explore AI Agent Development 🤖

📚 Table of Contents

  1. 📌 Overview
  2. 🛠️ Prerequisites
  3. 🚀 Create an Azure AI Foundry Project and Agent
  4. 🧠 Create Your Agent
  5. 🧪 Test Your Agent
  6. 🧹 Clean Up

📌 Overview

In this exercise, you will use the Azure AI Agent service in the Azure AI Foundry portal to create a simple AI agent that assists employees with expense claims.

🛠️ Prerequisites

  • An active Azure subscription
  • Access to the Azure AI Foundry portal
  • Optional but recommended: A resource group dedicated to AI exercises

🚀 Create an Azure AI Foundry Project and Agent

  1. Sign in to Azure AI Foundry
  2. Open your browser and go to:
    https://ai.azure.com
  3. Sign in with your Azure credentials.
  4. Close any tips or quick-start panes.

  5. If needed, click the Azure AI Foundry logo (top-left) to return to the Home page.

  6. Start Agent Creation

  7. On the Home page, click Create an agent.

  8. Configure Project

  9. Project name: Enter a valid name (e.g., ExpensesAgentProject).
  10. Expand Advanced options and fill in:
    • Azure AI Foundry resource: a valid resource name
    • Subscription: your Azure subscription
    • Resource group: select an existing group or create a new one
    • Region: any AI Services–supported region1

  11. Click Create and wait for provisioning.

  12. Deployment

  13. Once the project is created, the Agents playground opens automatically.
  14. A GPT-4o base model is deployed by default.

1 Some AI resources have regional model quotas. If you hit quota limits later, consider deploying in another region.

🧠 Create Your Agent

You will build a simple agent that answers questions based on a corporate expenses policy.

  1. Download the Policy Document 📄
  2. In a new browser tab, download the Contoso expenses policy:
    https://raw.githubusercontent.com/MicrosoftLearning/mslearn-ai-agents/main/Labfiles/01-agent-fundamentals/Expenses_Policy.docx

  3. Save it locally as Expenses_Policy.docx.

  4. Open the Setup Pane 🧩

  5. Return to the Agents playground tab.

  6. Locate the Setup pane (side or below the chat window).

  7. Agent Settings ⚙️

  8. Agent name: ExpensesAgent
  9. Model deployment: ensure your GPT-4o deployment is selected

  10. Instructions: click into the Instructions box and paste:

    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.

  11. Add Knowledge 📚

  12. Under Knowledge, click + Add → Files.
  13. In the dialog, create a new vector store named Expenses_Vector_Store and upload Expenses_Policy.docx.

  14. Verify that Expenses_Vector_Store appears in the Setup pane with 1 file.

  15. Add Actions 🧠💻

  16. Under Actions, click + Add → Code interpreter, then click Save.
  17. (No additional files are needed for the code interpreter.)

🧪 Test Your Agent

Use the playground chat to verify your agent’s behavior.

  1. Basic Query 💬
  2. Prompt: 
    What's the maximum I can claim for meals?

  3. The agent should answer based on the policy document.

  4. Submit a Claim – Part 1 📝

  5. Prompt: 
    I'd like to submit a claim for a meal.

  6. The agent should ask for your email address.

  7. Submit a Claim – Part 2 📧

  8. Provide an email (e.g., fred@contoso.com).

  9. The agent will then ask for a description and amount.

  10. Submit a Claim – Part 3 💲

  11. Prompt: 
    Breakfast cost me $20.

  12. The agent will use the code interpreter to generate a text file and provide a download link.

  13. Verify ✅

  14. Download the generated .txt file and open it to confirm the expense claim details.