Microsoft Foundry is Microsoft's platform for building AI applications and agents on Azure. It gives access to more than 10,000 models from Microsoft, OpenAI, Anthropic, Meta and others, and brings agents, models and tools under one Azure resource with shared identity, networking and policy.
Foundry Agent Service: three ways to build
- Prompt agents: the fastest path. Define instructions, a model and tools in the portal or SDK, and Foundry runs the agent with no code or infrastructure to manage.
- Voice-based prompt agents: managed real-time voice agents for contact-centre and phone scenarios, connected through Voice Live (some capabilities are in preview).
- Hosted agents: the most control. Bring your own code built with Agent Framework, LangGraph, the OpenAI or Anthropic agent SDKs, and Foundry runs it as a container with a managed endpoint, scaling and its own Entra identity.
Built for the enterprise
- Toolboxes: curate tools once (web search, file search, code interpreter, MCP servers, custom functions) and share them across agents through one governed endpoint.
- Observability: end-to-end tracing, evaluations and Application Insights, so you can see every decision an agent makes.
- Identity and security: Microsoft Entra identity, role-based access, content filters and virtual network isolation.
- Publishing and governance: publish agents to Microsoft Teams and Copilot, and manage them alongside Microsoft 365 agents through Microsoft Agent 365.
When to choose Foundry
Use Foundry when an agent needs your own data platform, custom models, complex reasoning or multi-agent orchestration, for example pricing and demand forecasting, investment screening or predictive maintenance. For process automation inside Microsoft 365, Copilot Studio is usually faster.
How Regulus can help
We design and build Foundry agents hosted in your GCC Azure region, connected to your data on Fabric or Databricks, and governed with Agent 365.