Skip to main content
Microsoft Foundry unifies agents, models, and tools under a single management grouping with built-in enterprise-readiness capabilities including tracing, monitoring, evaluations, and customizable enterprise setup configurations. The platform provides streamlined management through unified role-based access control (RBAC), networking, and policies under one Azure resource provider namespace.

​
Get started

Build your first agent in minutes, or open the portal to explore models and tools.

Build with models and agents

Open the Foundry portal | Get an Azure account

​
What you can build

​
Enterprise-ready platform

Foundry brings platform capabilities to every project. Some capabilities are in preview. For current status, see the general availability overview.
  • Observability: Trace and evaluate agents and models, and monitor them with built-in dashboards (preview). See Observability.
  • Governance and security: Apply Microsoft Entra identity, role-based access control, content filters, network isolation, and Azure Policy. See the Foundry control plane.
  • One management plane: Manage agents, models, and tools as a single Azure resource with unified access control, networking, and policies.

​
Start by building an agent

Most projects on Foundry center on an agent: a model paired with instructions and tools that can reason over a request and take action. The main decision is how much you want to customize and control how that agent runs. Think of it as a spectrum from declarative to full code.
  • Declarative, with the least to manage. Specify instructions, choose a model, and attach tools in the Foundry portal or with the SDK. Foundry hosts and runs the agent for you, with no application code or containers to maintain. In Foundry, this is a prompt agent. Create a prompt agent.
  • Full code, with the most control. Bring your own code or framework (for example, Microsoft Agent Framework, LangGraph, or Semantic Kernel), package it as a container, and Foundry runs it with a managed endpoint, scaling, identity, and observability. In Foundry, this is a hosted agent. Deploy a hosted agent.
You can start declarative and move to code as your needs grow. For a detailed comparison, see What are hosted agents?. For the end-to-end build, test, and ship cycle, see the agent development lifecycle. Not building an agent yet? If you only need to send prompts to a model with no tools or orchestration, start with a single model call.

​
Choose your developer surface

Foundry supports several surfaces. Many developers combine them, for example, prototyping in the portal and then moving to code. Follow these steps to go from zero to a working integration:
  1. Make your first model call. Set up your environment and send a prompt with the build with models and agents quickstart.
  2. Set up your developer environment. Install the CLI and SDK so you can build in code. See Set up your developer environment.
  3. Choose a model. Browse the catalog and compare options in the Foundry Models overview.
  4. Build your first agent. Start with a prompt agent, or go straight to a Hosted agent if you want to bring your own code.
  5. Add tools and knowledge. Extend your agent with tools, retrieval, and memory (preview). Start with the toolbox overview, the recommended way to add tools to an agent.

​
Evolution of Foundry

Foundry consolidates several previous Azure AI services and tools into a unified platform. For detailed guidance on transitioning, see Navigate from classic to the new experience.

​
Related content