OVERVIEW
Run production agents on a trusted, managed platform
Bring your preferred framework and model, then deploy with one command to a session-isolated managed runtime with built-in identity and observability so teams ship faster with less infrastructure glue.
- Move from local code to production with a session-isolated managed runtime. Build with the framework teams already use, deploy with one command, and run the model that fits the job—without rewrites or lock-in.
- Orchestrate multi-agent workflows, ground agents in enterprise knowledge with Foundry IQ (SharePoint, Fabric, Bing), connect to 1,400+ MCP-enabled tools, and use managed memory for persistent context without building and operating the plumbing.
- Run agents with Entra Agent ID, private networking options (including BYO VNet with no public egress), and flexible authentication for tool access then trace, evaluate, and enforce guardrails. Publish to Teams and Microsoft 365 Copilot in one click with tenant-wide governance.
FEATURES
Discover tools and capabilities to build your own agents
Hosted agents
Deploy with one command to hosted agents, a session-isolated managed runtime with autoscaling and zero idle cost.
Open standards
Keep tools and protocols open with MCP, A2A, and OpenAPI, so teams avoid protocol lock-in and can reuse integrations.
Built-in memory
Use managed short-term, long-term, and procedural memory so agents remember context without managing an external database.
Toolbox
Curate reusable tools and expose them through a single MCP-compatible endpoint, with centralized authentication, configuration, and versioned updates.
Voice Live integration
Enable natural, real-time speech-to-speech agent experiences with Voice Live integration for interactive, voice-first scenarios across enterprise applications.
Entra Agent ID integration
Give every agent a first-class identity with Entra Agent ID, access control, lifecycle management, and flexible authentication for tools.
Observability and guardrails
Gain visibility into agent quality with built-in evaluators, trace linkage, guardrails, and policy enforcement.
One-click publishing
Publish agents to Microsoft Teams and Microsoft 365 Copilot in one click, and manage visibility through tenant-wide governance.
Private networking
Run agents in your private network using bring-your-own virtual network (BYO VNet) and other private networking options to minimize public egress.
SECURITY
Built-in security and compliance
80K
Foundry is used by developers at more than 80,000 enterprises and digital natives, including 80% of Fortune 500 companies.
3B
daily enterprise search queries
11K+
Foundry Models to choose from—see why Microsoft Phi on Foundry Models has over 60 million downloads.
Pricing
Flexible, consumption-based pricing
Foundry Agent Service pricing is based on the individual models and tools you access and use from the service.
RELATED PRODUCTS
Azure products work better together
Build comprehensive solutions using Foundry Agent Service and other Azure AI products.
Find your AI models and tools
Discover Microsoft Foundry—your all-in-one platform for building transformative AI apps and agents.
CUSTOMER STORIES
Organizations are building custom agents with Foundry Agent Service
FAQ
Frequently asked questions
- Foundry Agent Service is a managed platform for building and running production AI agents on Azure. Bring the framework and model you choose, connect to enterprise knowledge and tools, and run with built-in identity, memory, and observability.
- Agents are your AI assistants that can help you optimize business processes and enhance productivity across all areas of your business. They are available in a wide array of capabilities, tools, and levels of complexity depending on your organization’s needs.
- AI agents are powered by several key capabilities that make them far more dynamic and versatile than traditional systems. For example, agents can use tools such as web search, code execution, and calendars, while retaining memory of past interactions to inform decisions. The ability to plan, act, and react in real time allows these agents to provide adaptive solutions for a variety of use cases. AI agents can even critique their own performance and collaborate with humans to improve their output.
- Foundry Agent Service addresses a wide range of use cases, including:
- Automating customer support by integrating with CRM systems to provide real-time assistance and resolve queries.
- Enhancing productivity by automating routine tasks and workflows, such as scheduling meetings, sending reminders, and generating reports.
- Improving decision-making by providing data-driven insights and recommendations based on real-time data analysis.
- Developers can easily build enterprise-grade agents that are secure, scalable, and performant. They can ground their agents with relevant, secure enterprise knowledge, automate complex workflows, and use prebuilt tools for advanced data analysis. Additionally, developers can choose the best model for their needs and benefit from scalable, secure memory management.
- Foundry Agent Service supports enterprise security patterns such as agent identity with Entra Agent ID, private networking options (including BYO VNet), and multiple authentication approaches for tool access. Teams can also use tracing to monitor and troubleshoot agent behavior.
- Hosted agents provide a managed runtime for running agent code in Azure. They help teams deploy agents without managing infrastructure and support running custom-code agents built with frameworks like Microsoft Agent Framework and LangGraph.
- Foundry Agent Service supports publishing agents to Microsoft Teams and Microsoft 365 Copilot. This helps teams bring agents to where users already work and manage distribution through existing Microsoft experiences.
Next steps
Choose the Azure account that’s right for you
Pay as you go or try Azure free for up to 30 days.
AI development tools
Design and manage AI applications
Create, customize, and scale AI apps and agents efficiently.
Products
Find the right AI products for your needs
Create the next generation of applications using artificial intelligence capabilities for any developer and any scenario.