Wait a minute - is it me or is OpenAI rolling out the managed Agents feature on the OpenAI Platform? Can anyone see it as well? 👀 Agent creation, environment creation, templates, and session creation now work. Sessions remain idle and throw an error for now.

Yessss, here it is 🙌

The orchestration and context recovery features are going to be game changers for developers. We actually went deeper on this here:

The templates give away the roadmap better than any announcement would — SRE incident response, bulk invoice and contract review, GitHub issue triage. Those are three headcount-shaped jobs, picked because they're repetitive and verifiable, and the sessions erroring out is the only reason this isn't launched yet.

the real product isn't agent creation. it's inspectable session state and clean recovery when a long run dies one step before the finish line.

Without EU residency its dead on arrival for any EU company. Its crazy to me that even the business plans dont have EU data residency, but just Enterprise. Anthropic doesnt offer anything. Only if you go through third party you might get EU servers.

agent creation working while sessions stay idle is the tell. is the error on session.create, or after the first tool call? that's whether the beta is a UI shell or the loop is actually up

Have something like that self-hosted. My version is build on terraform, making it stupidly easy to add everything you want. Python? Node? Java 25? Go? Rust? C++ with Unreal Engine? Last one will take some time to get ready though and requires GPU pass trough...

That matches the Agents API beta smell: create env/template/session works, then the long session dies on reconnect. The real question is whether the harness is Codex-backed cloud or a thin wrapper on local runs.