Anthropic’s biggest problem right now might not be competing with OpenAI or Google. It might be that its own users don’t know where to do anything. The company announced Claude Docs this week, a feature that lets Claude generate editable documents directly inside a chat window. But the more interesting part of this update is what it says about how fragmented the Claude product had become.
Claude Docs works roughly how you’d expect. Ask Claude to produce a one-page summary, a brief, or any structured document, and it drops the result into the conversation. The document is private by default, shareable with colleagues when you need it, and exportable to Google Docs or Microsoft Word. So if you’ve been copying Claude’s text output and pasting it somewhere else to format it, this cuts that step out.
Anthropic is also adding slide creation directly inside chat. Previously, if you wanted a presentation, you had to go to Claude Design, a separate section of the product. Now you can ask in a regular conversation, get the slides in-line, and either edit them there or download them as a PowerPoint or PDF. That’s a direct shot at workflows people currently handle with ChatGPT plus a separate tool, or with Google Gemini inside Workspace.
The bigger structural change is the merging of Cowork and Artifacts into a single interface. Cowork handled multi-step agentic tasks, things like reading your work files and generating documents or spreadsheets from them. Artifacts was a sidebar where Claude saved pieces of content for easy access. Both useful. But separate, with no continuity between them. Anthropic admitted that users were getting confused about which mode to use and that work started in one didn’t carry over to the other.
This matters because fragmentation is a real product liability at this stage of the AI assistant market. Microsoft Copilot has the Office integration advantage. Google Gemini lives inside tools people already use daily. Anthropic’s strength has been model quality, but if the interface requires users to mentally map which Claude mode handles which type of task, that quality gets buried under friction.
The new unified approach, which Anthropic is calling “One Claude,” is meant to let the model figure out what a task requires rather than making the user pick a mode first. These features are rolling out to Pro and Max subscribers over the next few weeks across web, desktop, and mobile, with Team and Free plan access coming later. Worth watching whether this consolidation actually simplifies the experience or just moves the confusion somewhere new.
