Love this end-to-end example from Jeff Hollan of what is now possible when you build long-running agents with Foundry. All now GA. Hosted Agents in Microsoft Foundry is now Generally Available! This brings agent native compute for your agents with any framework, any language, and any model 🚀 Build with the GitHub Copilot App, connect to Microsoft IQ, run in Foundry, securely interact with Microsoft Teams, govern in Agent 365, and optimize to compound your learnings continuously -- such a crazy valuable combo. Check out the demo / blog below to see it in action!
https://lnkd.in/gmdnrgNS The word doing the work here is long-running, and it changes the hard problem more than most people notice. A one-shot call either works or it does not, but an agent running for an hour stacks up small decisions, and a tiny error rate early quietly turns into a wrong outcome by the end. So the thing that actually matters stops being how smart the model is on any single step and becomes how well the system catches itself drifting mid-run. Teams that treat that reliability layer as the real product will trust agents with work that counts, while the ones assuming a capable model is enough will keep pulling agents back to supervised demos.
Amanda A.3d Building reliable, scalable, and continuously value-creating AI systems is a significant step forward. As these agent technologies mature, will they bring automation, decision support, or entirely new workflows?
The shift toward 'agent-native compute' is a game-changer for building persistent, long-running systems. By enabling agents to work across frameworks and models while continuously compounding learnings within Foundry, Microsoft is providing the infrastructure necessary to move from static automation to truly dynamic, self-improving workflows. This is the structural evolution required to scale intelligence at the enterprise level." I'm designing the ARO project with a similar focus on continuous learning loops for student growth, and seeing this architectural model reach this level of maturity is highly motivating."
Your example highlights how AI is evolving from simple assistants to autonomous agents capable of managing complex, long-running workflows. This shift isn't just about increasing productivity; it's about fundamentally changing how organizations design, execute, and scale business processes. In industries like travel technology, autonomous AI agents have the potential to transform booking operations, customer support, revenue management, and back-office automation. The competitive advantage will belong to organizations that redesign their workflows around AI rather than simply adding AI to existing processes. Which business function do you believe will experience the greatest transformation through autonomous AI agents over the next three years? The next wave of digital transformation will be led by businesses that combine AI agents with strong business strategy and human oversight.
The next bottleneck may not be agent capability but organizational observability. As agents begin operating for hours or days, success will depend less on individual model quality and more on whether organizations can continuously measure goal alignment, confidence drift, context integrity, resource efficiency, and recovery after failure. Long-running agents donot just require better AI. They require better operational science. That's where the next competitive advantage is likely to emerge. 🚀
Stacy Budd2d Congrats Jeff and the Foundry team! This is a solid milestone. Long-running agents that integrate so well with GitHub Copilot, Teams, Agent 365, and can continuously learn that combo is really powerful. The end-to-end example looks promising. Excited to see how people start using it
Arda Arık3d The real shift is from AI that responds to AI that persists, coordinates, and acts. Long-running agents could become the operational layer that turns intelligence into continuous business execution.
Brendon Geils23h The infrastructure layer for hosting agents is commoditizing fast, which is the right move for adoption. The differentiation shifts to who controls the enterprise workflows and the data these agents operate on, and that's where the next wave of competitive advantage gets built. How quickly do you see enterprises moving from experimentation to production-scale agent deployments?
Edwin Uchman3d This is where AI starts moving from “chat” to real operations. Long-running agents are not just another productivity feature. They can become a new execution layer inside the company. Research. Analysis. Documentation. Workflow coordination. Customer follow-up. Internal reporting. Market monitoring. Process improvement. But this also creates a serious leadership question: Is the organization ready to delegate work to agents responsibly? Because agents will not fix unclear ownership. They will expose it. They will not fix messy processes. They will run into them faster. They will not fix weak communication. They will amplify the consequences. That is why enterprise AI is no longer only an IT topic. It is an operational maturity test. The companies that win will not be the ones that simply “use agents.” They will be the ones that redesign work around clarity, accountability, clean data, and faster execution. Edwin Uchman I THE MARGIN HEMORRHAGE
Sandra Kane2d This is huge! Congrats to Jeff and the whole team. Long-running agents that actually integrate across Copilot, Teams, M365, and Agent 365 feel like a real leap forward. The end-to-end workflow looks super smooth. Definitely going to check out that demo. Well done!