Agent frameworks solved one problem. What solves the next one?
这条记录涉及编程工具或代码能力更新，适合开发者评估工作流变化和可复用价值。
过去一年，代理框架、编排库和编码代理激增，代理的创建变得越来越容易。然而，随着组织开始在不同团队、工作流和环境中运行数十甚至数百个代理，挑战将从代理创建转向代理运营。部署、访问控制、治理、可观测性、评估、审计跟踪、版本控制和生命周期管理等问题将变得比提示工程更为重要。目前，许多公司拥有代理，但缺乏有效的管理方式。…
Over the last year we've seen an explosion of agent frameworks, orchestration libraries, and coding agents. Building agents is becoming easier every month, and honestly, that's no longer the part I find most interesting. The bigger question is what happens after an organization starts running dozens or hundreds of agents across different teams, workflows, and environments. At that point, the challenge stops being agent creation and starts becoming agent operations.
Things like deployment, access control, governance, observability, evaluation, audit trails, versioning, and lifecycle management start looking a lot more important than prompt engineering. It almost feels like the ecosystem is heading toward a world where every company has agents, but very few have a good way to manage them. Makes me wonder whether the agent control plane will become a real category over the next few years, similar to how Kubernetes emerged once containers became mainstream..