Skip to main content

Welcome to Azure AI Documentation

Azure AI is Microsoft’s comprehensive cloud platform for building, deploying, and managing enterprise AI applications. From cutting-edge agents and large language models to production-ready machine learning pipelines and intelligent search, Azure AI provides everything you need to innovate with confidence.

Microsoft Foundry

Unified platform for building AI agents, working with models, and deploying generative AI applications at enterprise scale.

Azure Machine Learning

End-to-end platform for training, deploying, and managing machine learning models with MLOps best practices.

Azure AI Search

AI-powered information retrieval platform for building rich search experiences and RAG applications.

Open Datasets

Curated public datasets for enriching machine learning solutions with weather, census, and location data.

Why Azure AI?

Azure AI brings together the best of Microsoft’s AI capabilities under one unified platform:

Enterprise Ready

Built-in security, compliance, and governance features for production workloads.

Integrated Platform

Seamless connections between services, from data ingestion to model deployment.

Azure Scale

Global infrastructure with reliability, monitoring, and automatic scaling.

Getting Started

New to Azure AI? Follow our quickstart guide to create your first project and deploy an AI solution.
1

Choose Your Service

Select the Azure AI service that best fits your needs - agents, ML models, search, or data enrichment.
2

Create Resources

Set up your Azure account and create the necessary workspace or project resources.
3

Build & Deploy

Use the portal, SDKs, or APIs to build and deploy your AI solution.
4

Monitor & Optimize

Track performance, evaluate results, and continuously improve your models.

Key Capabilities

Multi-Agent Orchestration

Build sophisticated AI agents that collaborate to solve complex problems. Microsoft Foundry provides SDKs for Python and C# that enable advanced workflow execution and multi-agent coordination.

Model Flexibility

Work across model providers with a consistent API contract. Access Azure OpenAI, models from the Foundry model catalog including Mistral, Meta, Cohere, and more.

Production MLOps

Streamline the machine learning lifecycle with automated pipelines, model versioning, monitoring, and deployment management. Azure Machine Learning provides enterprise-grade MLOps capabilities.

Intelligent Retrieval

Combine classic search with modern agentic retrieval for grounded, context-aware responses. Azure AI Search supports full-text, vector, hybrid, and multimodal search.

Platform Features

  • Microsoft Entra ID authentication
  • Azure Virtual Networks with network security groups
  • Azure Key Vault for secrets management
  • Role-based access control (RBAC)
  • Document-level access control
  • Azure Policy integration
  • SDKs for Python, C#, JavaScript/TypeScript, and Java
  • VS Code extensions for local development
  • REST APIs for automation
  • Jupyter notebooks in the cloud
  • Azure CLI integration
  • Real-time performance metrics
  • Built-in tracing and debugging
  • Azure Monitor integration
  • Custom evaluation frameworks
  • Cost tracking and optimization

Explore Services

Each Azure AI service is designed to solve specific challenges while working together as part of a unified platform:

Microsoft Foundry

A unified Azure platform-as-a-service offering for enterprise AI operations, model builders, and application development. Build generative AI applications and AI agents with cutting-edge models, grounded in responsible AI practices. Key Features:
  • Foundry Agent Service for building intelligent agents
  • Multi-agent orchestration and workflows
  • Access to 1,400+ tools through the tool catalog
  • Memory capabilities for contextual interactions
  • Knowledge integration with Foundry IQ
  • Real-time observability and monitoring

Azure Machine Learning

Accelerate and manage the complete machine learning project lifecycle. Train and deploy models at scale with built-in MLOps, supporting frameworks like PyTorch, TensorFlow, and scikit-learn. Key Features:
  • Automated ML for rapid model development
  • Distributed training on GPU clusters
  • Managed endpoints for real-time and batch scoring
  • Model catalog with LLMs and foundation models
  • Prompt flow for LLM application development
  • Integration with Azure Synapse and Databricks
Fully managed cloud search service that connects data to AI. Unifies access to enterprise and web content for agents and LLMs to produce reliable, grounded answers. Key Features:
  • Classic search and agentic retrieval modes
  • Full-text, vector, hybrid, and multimodal search
  • AI enrichment with chunking and vectorization
  • Knowledge bases for multi-source retrieval
  • Integration with Azure OpenAI and Foundry
  • Document-level security and compliance

Azure Open Datasets

Curated public datasets optimized for machine learning workflows. Enrich your models with weather, census, transportation, and economic data. Key Features:
  • Pre-processed and cleaned datasets
  • Regular updates from authoritative sources
  • Integration with Azure ML and Databricks
  • Python SDK for easy access
  • Transportation, labor, and weather datasets

Next Steps

Ready to start building with Azure AI?

Quickstart Guide

Get up and running with Azure AI in minutes

Services Overview

Explore all available Azure AI services

Tutorials

Step-by-step guides for common scenarios

API Reference

Complete API documentation and SDKs
New to Azure? Create a free Azure account to get started with credits for Azure AI services.

Community & Support

Join the Azure AI community and get help when you need it:
  • Documentation: Comprehensive guides and references
  • Microsoft Learn: Free training and certification paths
  • Stack Overflow: Community Q&A with the azure-ai tag
  • GitHub: Sample code and open-source projects
  • Azure Support: Professional support plans available

Pro Tip: Start with Microsoft Foundry if you’re building agent-based applications, or Azure Machine Learning if you’re focusing on custom model training and MLOps.