Anthropic integrates Cowork into Claude

Merging Cowork into core Claude shows that Anthropic no longer wants agent work to feel like a separate expert mode, it wants long running automation to become the default way people use Claude. That matters because the main friction in early Cowork was not model quality alone, it was switching surfaces, setting up tools, and managing technical workflow steps. Folding delegation, progress checking, files, and output creation into one thread makes Claude feel less like a chatbot plus sidecar and more like a work operating system.1 2 3 4

• Early Cowork usage was concentrated among power users who could debug broken multi tool chains and navigate technical setup. One interview found roughly 15% of users generated 75% to 80% of output, with adoption stalling when Claude exposed GitHub and permissions details to non technical users.1
• The product pattern is the same one showing up across the market. OpenAI is expanding from chat into Codex and work agents, while tools like Perplexity Computer, Replit, and Glean are all collapsing search, generation, actions, and artifact creation into one workspace instead of separate apps.3 4 5
• Anthropic has been moving toward work rather than pure consumer chat for a while. Earlier positioning already emphasized Projects, enterprise controls, and document grounded workflows, so unifying Cowork with chat is a continuation of that strategy, not a side experiment.2 3

The next step is a Claude that hides orchestration almost completely. The winning assistant will not ask users to choose between chat, agent, docs, slides, or design. It will take one request, pull the right context, run the work in the background, and return something editable and ready to share. That is where the market is heading, and Anthropic is pushing Claude directly into that race.1 4 5