Anthropic announced on Tuesday that Claude Cowork, its AI agent for performing digital tasks, is expanding beyond the desktop app to iOS, web browsers, and cloud-based persistent operation. The update eliminates the most common complaint about Cowork since its January launch: users had to keep a laptop open and the desktop app running for the agent to stay active, according to WIRED.
What Changed
Previously, Claude Cowork’s Dispatch feature let users send task requests from their phone, but the desktop session had to remain active for the agent to execute. Users who wanted overnight task completion resorted to leaving laptops half-open. The new release removes that dependency entirely: Cowork now runs autonomously in the cloud regardless of device state, with smartphone push notifications and approval prompts delivered via the Claude iOS app, WIRED reported.
The update also merges the Claude chatbot interface with the Cowork agent in browser and desktop versions, creating a unified experience rather than separate tools.
90% of Usage Is Not Coding
Alongside the release, Anthropic published internal usage data showing that 90% of Claude Cowork sessions involve non-coding tasks, according to ZDNET. The two largest usage categories are “business process and operations” (data reports, checklists) and “content creation and copywriting” (slide decks, partnership proposals), as WIRED reported.
That figure positions Cowork as a general-purpose productivity agent rather than a developer tool, placing it in direct competition with always-on agent platforms like OpenClaw that already serve non-technical users.
Pricing and Rollout
The revamped Cowork launches as a beta for Max plan subscribers ($100/month), with features expected to trickle down to Pro ($20/month) subscribers. Free-tier users do not currently have access to Claude Cowork, WIRED noted.
The Competitive Landscape
The expansion follows a pattern that started with OpenClaw’s viral adoption in early 2026. OpenAI responded by hiring OpenClaw’s creator and launching Codex; Google launched Spark; and now Anthropic is removing the last friction point from its own agent offering. OpenAI rolled out Codex Remote in June with similar smartphone-control capabilities, and added iOS-native Codex task management in July, WIRED reported.
The strategy across all three companies converges on the same bet: agentic automation belongs inside the chatbots millions of people already have on their phones, not in standalone developer tools.