PaPoo
cover

Claude Stops Making You Pick a Mode

The part that caught my eye is not the new Docs or Slides bits. It’s Anthropic admitting, implicitly, that the Chat/Cowork split was friction. That sounds obvious in hindsight, but product teams often keep these splits around far too long because they feel neat in a demo. Real users just feel like they’re being asked to guess which door to walk through.

I’d actually trust this change more than I usually trust “unification” announcements, because the article says existing chats, projects, artifacts, connectors, and skills carry over. That’s the difference between a useful consolidation and a cosmetic rename. If Claude really can start with a quick question and then keep going into a larger task without forcing me to re-orient the whole workspace, that’s a better shape for an assistant.

What I’m less certain about is the claim that Claude can now “judge” what tool or skill to use and route tasks accordingly. That may be true in a narrow sense, but it’s also where these systems like to sound more autonomous than they are. In practice, the quality of the handoff matters more than the slogan. If it picks the wrong path, or asks for confirmation too often, the elegance disappears fast.

The Docs and Slides angle is more interesting to me than the branding cleanup. They’re basically saying: stop thinking in terms of separate modes, and let a conversation branch into document production when needed. That is a sensible direction, especially if the output can be edited in place and exported cleanly to PowerPoint or PDF. But I’d still want to see how good the generated decks actually are before getting excited. Most “AI presentation” tools are great at producing something that looks finished and less great at producing something I’d be willing to send to a client.

The manual/auto split is the detail I’d watch. Letting Claude keep going without step-by-step permission sounds useful, but only if the safety checks are conservative enough to avoid ugly surprises. The article frames this as a user choice, which is right. I think the real question is whether the auto mode feels like a productivity boost or just a way to make the assistant more confident while quietly increasing the blast radius when it’s wrong.


Reference: 「Claude」のチャットと「Cowork」が1つに。「Claude Docs」「Claude Slides」も登場/「Pro」「Max」プランから順次展開。「Free」プランにも近く提供される予定

同じ著者の記事