How to get started with Azure AI Foundry SDK

This article provides an overview of the Azure AI Foundry SDK and how to get started using it.

Azure AI Foundry SDK client libraries

The Azure AI Foundry SDK is a comprehensive toolchain designed to simplify the development of AI applications on Azure. It enables developers to:

• Access popular models from various model providers through a single interface
• Easily combine together models, data, and AI services to build AI-powered applications
• Evaluate, debug, and improve application quality & safety across development, testing, and production environments

The Azure AI Foundry SDK is a set of client libraries and services designed to work together.

This article applies to a Foundry Project. For more information, see Types of projects.

Prerequisites

• An Azure subscription. If you don’t have one, create a free account.

• Create a Foundry project if you don’t have one already.

• Sign in with the Azure CLI using the same account that you use to access your project:

  az login
  

Unified Projects client library

The Azure AI Foundry Projects client library is a unified library that enables you to use multiple client libraries together by connecting to a single project endpoint.

Python

• Install the project client library

  pip install azure-ai-projects azure-identity
  

• Create a project client in code. Copy the Azure AI Foundry project endpoint from the Overview page of the project and update the connections string value.

  from azure.identity import DefaultAzureCredential
  from azure.ai.projects import AIProjectClient
  
  project = AIProjectClient(
    endpoint="your_project_endpoint",  # Replace with your endpoint
    credential=DefaultAzureCredential())
  

Java

• Add these packages to your installation (preview):

  • com.azure.ai.projects
  • com.azure.core

• Create a project client in code. Copy the Azure AI Foundry project endpoint from the Overview page of the project and update the connections string value.

  import com.azure.ai.projects.ProjectsClient;
  import com.azure.ai.projects.ProjectsClientBuilder;
  import com.azure.core.credential.AzureKeyCredential;
  
  String endpoint ="your_project_endpoint"; // Replace with your endpoint
  
  ProjectsClient projectClient = new ProjectsClientBuilder()
      .credential(new DefaultAzureCredential())
      .endpoint(endpoint)
      .buildClient();
  

JavaScript

• Install dependencies (preview):

  npm install @azure/ai-projects @azure/identity
  

• Create a project client in code. Copy the Azure AI Foundry project endpoint from the Overview page of the project and update the connections string value.

  import { AIProjectClient } from '@azure/ai-projects';
  import { DefaultAzureCredential } from '@azure/identity';
  
  const endpoint = "your_project_endpoint"; // Replace with your actual endpoint
  const project = new AIProjectClient(endpoint, new DefaultAzureCredential());
  

C#

• Install packages:

  dotnet add package Azure.Identity
  dotnet add package Azure.Core
  dotnet add package Azure.AI.Inference
  

• Create a project client in code. Copy the Azure AI Foundry project endpoint from the Overview page of the project and update the connections string value.

  using Azure;
  using Azure.Identity;
  using Azure.AI.Inference;
  using Azure.Core;
  using Azure.Core.Pipeline;
  
  var endpointUrl = "your_project_endpoint"; // Replace with your actual endpoint
  var credential = new DefaultAzureCredential();
  
  AzureAIInferenceClientOptions clientOptions = new AzureAIInferenceClientOptions();
  BearerTokenAuthenticationPolicy tokenPolicy = new BearerTokenAuthenticationPolicy(
      credential, 
      new string[] { "https://cognitiveservices.azure.com/.default" }
  );
  clientOptions.AddPolicy(tokenPolicy, HttpPipelinePosition.PerRetry);
  
  var projectClient = new ChatCompletionsClient(
      endpointUrl, 
      credential,
      clientOptions
  );
  

• Using the project endpoint, you can:
  • Use Foundry Model, including Azure OpenAI
  • Use Foundry Agent Service
  • Run evaluations in the cloud
  • Enable tracing for your app
  • Retrieve endpoints and keys for external resource connections