Azure AI Foundry
by Microsoft
The unified platform to build, deploy, and govern enterprise AI apps and agents
Azure AI Foundry (formerly Azure AI Studio, now also branded Microsoft Foundry) is Microsoft's unified enterprise platform for building, customizing, deploying, and governing generative AI applications and agents. It combines model access, an agent service, tooling, and enterprise-grade security and observability in a single portal.
At a Glance
- Category
- Enterprise Platform
- Pricing
- Usage-based, Pay-as-you-go
- Target Market
- CIOs, CTOs, Enterprise Developers, Data Scientists, AI/ML Engineers, IT Decision Makers
- Founded
- 2024
- Headquarters
- Redmond, Washington, United States
Key Features
- ✓Model Catalog
Unified access to OpenAI, Microsoft (MAI, Phi), xAI Grok, DeepSeek, Anthropic Claude, and open-source models in one portal.
- ✓Foundry Agent Service
Generally available agent runtime with Model Context Protocol support and multi-agent orchestration via a unified Semantic Kernel and AutoGen framework.
- ✓Foundry IQ
Agentic RAG engine powered by Azure AI Search that grounds agents on enterprise sources like SharePoint and ADLS with Purview policy enforcement.
- ✓Foundry Control Plane
Centralized governance, security, and observability layer with Microsoft Defender, Purview, and Entra ID integration.
- ✓Fine-tuning and distillation
Model customization, fine-tuning, and distillation through integrated Azure Machine Learning tooling.
- ✓Foundry Local
Run models and AI capabilities on-device and at the edge for hybrid and offline scenarios.
Capabilities
Use Cases
- •Enterprise AI agents
Build, evaluate, and deploy multi-agent workflows that securely access internal systems and data.
- •Retrieval-augmented chat over enterprise data
Ground generative AI responses on private SharePoint, web, and storage sources with enforced security policies.
- •Custom and fine-tuned model deployment
Customize frontier and open models for domain-specific tasks and deploy them with managed infrastructure.
Ideal For
Best For
- ✓Building and orchestrating enterprise AI agents
- ✓Accessing and customizing OpenAI and other frontier models in a governed environment
- ✓Deploying generative AI apps with enterprise security and compliance
Integrations
Deployment
Market Analysis
Pros
- ✓Broad, unified model catalog including exclusive OpenAI models
- ✓Strong enterprise security, governance, and compliance breadth
- ✓Seamless integration with Microsoft 365 and Azure ecosystem
- ✓GA agent service with MCP and multi-agent orchestration
Cons
- ✗Pricing across compute, tokens, and services can be complex to estimate
- ✗Best value realized within a Microsoft-centric stack
- ✗Early iterations were criticized for rough documentation and reliability
Pricing
Pay-as-you-go
Usage-based
- ✓Per-token model inference
- ✓Compute and storage billed by consumption
- ✓Access to model catalog, agents, and tools
Provisioned Throughput (PTU)
Contact for pricing
- ✓Reserved provisioned throughput units
- ✓Predictable performance for steady workloads
- ✓Hourly or monthly reservation pricing
Managed Compute
Usage-based
- ✓Dedicated GPU infrastructure (A100, H100, H200, MI300)
- ✓Deploy open-source and custom models
- ✓Autoscaling, pay only for what you use
Billing is consumption-based across model inference (per 1M tokens), compute, storage, and add-on services such as Content Safety and Azure AI Search; provisioned throughput and reservations are available for predictable workloads. Try Azure for free is offered for new accounts.
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Writer
The full-stack generative AI platform for the enterprise
Google Vertex AI
Google Cloud's unified platform to build, deploy, and scale machine learning and generative AI
Dataiku
The Universal AI Platform for analytics, machine learning, and AI agents.
DataRobot
The enterprise AI platform to build, operate, and govern AI agents and applications at scale.