Azure AI Agent Service Lab

What is Azure AI Agent Service

Azure AI Agent Service is a fully managed platform designed to help developers create, deploy, and scale AI agents with ease, without the need to manage underlying compute and storage resources. By utilising advanced language models, it simplifies the process of building AI agents that can answer questions, perform tasks, and automate workflows. The service offers seamless integration with various tools and data sources, enabling agents to access and interact with real-world data. It provides features like automatic tool calling, secure data management, and extensive data integrations, making it ideal for enterprise applications requiring enhanced security and flexibility.

Table of Contents

1. What is Azure AI Agent Service
2. Azure AI Agent Service vs. OpenAI Assistants API
3. Concepts
4. Technical Prerequisites
5. Tutorial Series

Azure AI Agent Service vs. OpenAI Assistants API

Feature	Azure AI Agent Service	OpenAI Assistants API
🧠 Model Selection	OpenAI, Llama 3, Mistral, and Cohere	OpenAI
📚 Knowledge	Microsoft Fabric, SharePoint, Bing Search, Azure AI Search, and File Search	File search
🛠️ Tools	Azure Logic Apps, OpenAPI, Azure Functions, Code Interpreter, and Function Calling	Code Interpreter and Function Calling

Concepts

Concept	Explanation
🤖 Agent	An AI-driven microservice that uses models and tools to perform tasks, answer questions, and automate workflows. Think of it as a virtual assistant powered by AI.
🧵 Thread	A conversation session between an agent and a user. It stores messages and manages conversation history, ensuring the content fits within the model's context.
✉️ Message	A piece of communication within a thread, which can be text, images, or other files, created by either the agent or the user.
🚀 Run	The activation of an agent to perform tasks based on the thread's messages. During a run, the agent processes the information and may call models and tools to achieve its objectives.
📝 Run Step	A detailed list of actions taken by the agent during a run. Each step shows how the agent interacts with tools, models, and messages to reach its final outcome. Examining run steps helps you understand the agent's decision-making process.

🤔 Technical Prerequisites

1. Azure Subscription:

   • You need an active Azure subscription. You can create one for free if you don't have it.

2. Python Environment:

   • Install Python 3.8 or later. Ensure your Python environment is set up and properly configured.

3. Azure AI Developer Role:

   • Assign the Azure AI Developer RBAC role at the Hub or Project level to access and manage Azure AI services.

4. Azure CLI:

   • Install the Azure CLI. It is used to log in to your Azure subscription.

5. Clone the Repository:

   • Use Git to clone the repository containing the tutorials to your local machine:

     git clone https://github.com/tayganr/azure-ai-agent-lab.git

   • Navigate into the cloned repository directory:

     cd azure-ai-agent-lab

6. Python Virtual Environment:

   • Create a Python virtual environment to manage dependencies cleanly:

     python -m venv myenv  

   • Activate the virtual environment:
     • On Windows:

       myenv\Scripts\activate  

     • On macOS/Linux:

       source myenv/bin/activate  

7. Python Packages:

   • Install necessary Python packages using pip and a requirements file:

     pip install -r requirements.txt  

8. Connection String:

   • Create a connection string using information from your Azure AI project. This includes HostName, AzureSubscriptionId, ResourceGroup, and ProjectName. Set this as an environment variable PROJECT_CONNECTION_STRING:

     • For PowerShell:

       $env:PROJECT_CONNECTION_STRING="your-connection-string-here"  

     • For Bash (macOS/Linux):

       export PROJECT_CONNECTION_STRING="your-connection-string-here"  

     • For Windows Command Prompt:

       set PROJECT_CONNECTION_STRING="your-connection-string-here"  

   • You can obtain the connection string from the Azure AI Foundry portal.

9. Login to Azure:

   • Use the Azure CLI to log in to your Azure subscription:

     az login  

By completing these steps, you'll be set up to run through the tutorials in the repository, enabling you to explore and learn about the Azure AI Agent Service.

📚 Tutorial Series

To help you get started with the Azure AI Agent Service, we've prepared a series of tutorials. These tutorials will guide you through the process of building and enhancing AI agents using the service.

1. Getting Started with Azure AI Agents
   Learn how to create a basic agent using the Python SDK, focusing on instructions and messages.

2. Integrating File Search
   Discover how to add file search capabilities to your agent, enabling it to handle file-based interactions.

3. Grounding Your Agent with Bing Search
   Integrate Bing Search to fetch real-time web data and enhance agent responses.

4. Implementing the Code Interpreter Tool
   Utilize the Code Interpreter tool to perform complex computations within your agent.

5. Creating a Multi-Tool Agent
   Build a comprehensive agent solution that leverages multiple tools for advanced functionality.

Feel free to follow these tutorials in order to gradually build up your skills and understanding of the Azure AI Agent Service.