Claude, this is lifting the table again. What you see is the squeeze on other agents, and what I see is the leaders crazily seizing the future entrance to AI agents. Seeing Claude announce the launch of Cluade Managed Agents, a managed AI agent platform in the early hours of the morning, the threshold for "AI agents from prototype to production deployment" has been significantly lowered, allowing users or customers to quickly put agents online and run, without the need to spend a long time building infrastructure as before. Before that, if you wanted to be a proxy that could truly run in the business (such as automatically handling Notion tasks, being an AI teammate in Asana, and automatically fixing bugs and opening PRs in Sentry), you had to do your own orchestration loops, tool calls, various tests, and various monitoring (logging, scaling, upgrading), which was time-consuming and laborious. Now Claude Managed Agents has completely taken care of these "dirty and tiring tasks", and users only need to define: tasks+tools+rules. Simply put, users only need to determine what to do, which tools to use, and roughly how to do it. This hosted agent platform will automatically set up and provide one click full package for you. This is what Anthropic has taken one step further from "selling only models (LLM)" to directly providing the "hosted agent running platform" end-to-end to users, truly turning "AI agents" from toys into productivity tools. For Authropic himself: Claude cowork enables it to further integrate with AI agents from model suppliers; Now Cluade Managed Agents has further become a managed agent platform. This will ultimately lead to higher platform stickiness, higher willingness of enterprises to pay, and stronger long-term control - in the future, when enterprises use proxies, the place where they spend money is no longer just tokens, but the entire hosting service. Seeing that many people's comments still focus on squeezing other agents, we actually need to take a closer look. Just like the last time Openclaw was crazily banned, in essence, we still need to seize the "entry effect" of AI agents, so that users can think of Claude first when using agents, and directly use Claude's entry instead of other platforms' entry. Authenticic is doing everything possible to avoid becoming a token provider. Today's move is still a competition for access to AI agents. Of course, I believe other countries will also see this and probably accelerate to catch up. Intense competition