Driving Stop-and-Go Business Processes to Closure with Foundry Hosted Agents

2026-06-08 · Source: Microsoft Foundry Blog articles · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure · Depth: Intermediate, medium

Summary

Microsoft Foundry Hosted Agents provide a solution for building stateful AI agents capable of managing long-running business processes, such as Statement of Work (SOW) response orchestration, which can span days or weeks. The core challenge of stateless agent loops is overcome by leveraging Foundry's per-project sandbox microVMs to persist agent state. This architecture ensures that the agent reconstructs its complete operational context from disk upon each wake-up, eliminating the need for external databases and keeping the client application thin. Built using the Microsoft Agent Framework (MAF) and integrating OpenAI models on Foundry, these agents connect to Microsoft 365 via WorkIQ MCP endpoints and offer turnkey OpenTelemetry tracing for comprehensive monitoring.

Key takeaway

For AI Architects designing agents for complex, multi-day business workflows, consider Microsoft Foundry Hosted Agents. This approach centralizes agent state within per-project microVMs, eliminating the need for external databases and enhancing data residency compliance. Your teams can achieve robust, scalable, and collaborative agent solutions, especially in regulated environments, by adopting this server-side state management pattern over local client-side alternatives.

Key insights

Foundry microVMs enable stateful AI agents for long-running business processes without external databases.

Principles

• Agent state lives in the sandbox, not the client.
• Reconstruct full agent context on every wake-up.
• Client remains thin, holding only session IDs.

Method

Mint a unique "agent_session_id" for each project, linking it to a dedicated microVM. Embed this ID in all requests to reattach the sandbox, and use "state_patch_json" for atomic state updates.

In practice

• Use "agent_session_id" to persist agent state.
• Implement "state_patch_json" for granular updates.
• Integrate OpenTelemetry for agent loop tracing.

Topics

• Foundry Hosted Agents
• Stateful AI
• MicroVMs
• Microsoft Agent Framework
• SOW Orchestration
• OpenTelemetry
• M365 Integration

Code references

• MSFT-Innovation-Hub-India/charter-agent

Best for: AI Engineer, MLOps Engineer, AI Architect

Related on AIssential

• How Microsoft Used Its Own Tech To Lead The Gartner® Magic Quadrant™ for AI Application Development — Microsoft Foundry and its Agent Framework streamline multi-agent AI application development from ideation to production.
• Bring Your Own Model to Foundry Agent Service Is Now Generally Available — Foundry Agent Service's BYOM feature allows enterprises to integrate custom or third-party AI models via their own gateways.
• Every Agent Needs a Box — Aaron Levie, Box — Secure, sandboxed environments are crucial for AI agents to operate effectively and safely within enterprise data systems.
• Microsoft's AI Futurist explains how he uses Copilot — and the real-world problems enterprises are solving with agents — Microsoft's agent strategy centers on providing a comprehensive platform for context, governance, identity, and secure data access, alongside diverse model choices.
• Custom Agents now available on Databricks — Databricks Custom Agents streamline AI agent development and deployment with serverless infrastructure and integrated governance.
• Long-running Agents — Long-running agents overcome AI limitations by externalizing state, enabling sustained, multi-session progress on complex tasks.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Microsoft Foundry Blog articles.