OFICIAL Microsoft Source

Jeff Hollan on Building Long-Running Agents with Foundry

What happened
Based on Microsoft Source · Jul 10, 2026

Microsoft Foundry now offers generally available hosted agents, enabling developers to build long-running AI agents across frameworks, languages, and models with native compute and continuous optimization.

🚩 Report a correction ↗ Open source
Key points
·
Love this end-to-end example from Jeff Hollan of what is now possible when you build long-running agents with Foundry.
·
Hosted Agents in Microsoft Foundry is now Generally Available!
·
The combination of building via GitHub Copilot, running securely on Foundry, and gaining continuous optimization through Microsoft IQ makes it incredibly seamless to move from a basic prompt to a robust, long-running autonomous worker.
·
The word doing the work here is long-running, and it changes the hard problem more than most people notice.

Microsoft Foundry’s Hosted Agents, now generally available, provide developers with a production-ready runtime for AI agents built using any framework, language, or model. This eliminates infrastructure bottlenecks by offering agent-native compute, allowing teams to transition from basic prompts to robust, long-running autonomous systems without managing underlying infrastructure.

The integration with GitHub Copilot, Microsoft IQ, Teams, and Agent 365 enables continuous learning and optimization, transforming AI from simple assistants into persistent, self-improving agents. This shift moves beyond one-shot interactions, emphasizing system reliability to prevent small errors from compounding into significant failures over extended operations.

By enabling agents to work across frameworks and models while continuously compounding learnings within Foundry, Microsoft is facilitating a shift from static automation to dynamic, self-improving workflows. This structural evolution supports enterprise-scale intelligence by providing the infrastructure needed for persistent, long-running systems.

The release underscores a broader trend toward operational maturity in enterprise AI, where organizations must redesign workflows around clarity, accountability, and clean data. Long-running agents expose inefficiencies in processes and ownership, making them a test of organizational readiness rather than just an IT implementation.

Original source → Deals on Clipraptor.com →