Anthropic Expands Claude Cowork Agent to Mobile and Web, No Desktop Required
Summary
- • Claude Cowork now runs on mobile and web, eliminating need to keep laptop active
- • Available to Max subscribers ($100/month) with Pro tier rollout planned next
- • Usage data: business operations 33.4%, content creation 16.4%, coding just 8.7% of sessions
- • Part of broader industry race toward always-on background AI agents across devices
Details
Mobile and web expansion
Claude Cowork launched as desktop-only in January 2026; as of July 2026 it is available on web and mobile for Max plan subscribers at $100/month, with Pro ($20/month) rollout planned.
Background task execution
Tasks now run without an active desktop session — Cowork can execute multi-hour workflows overnight and push updates to mobile, replacing the previous Dispatch feature that required a woken machine.
Usage breakdown
Business process & operations: 33.4%; content creation & copywriting: 16.4%; software development: 8.7% — based on 1.2M anonymized sessions from 600K+ organizations in late May 2026.
Non-developer targeting
Anthropic explicitly targets roles in finance, HR, administration, marketing, and management — signaling a strategic push beyond developers into general knowledge work.
Competitive landscape
OpenAI (Codex), Google (Spark), and Anthropic (Cowork) are racing to build always-on cross-device agents — a trend sparked by the viral homebrew agent OpenClaw at the start of 2026.
Claude Tag for Slack
Anthropic also launched Claude Tag, an always-on Claude integration in Slack that acts as an AI teammate, further extending agent presence into collaboration tools.
Product updates and usage data for Anthropic's Claude Cowork mobile and web expansion.
What This Means
Cowork's mobile launch signals Anthropic is competing for something larger than the AI assistant market: it wants to own the multi-device, always-on layer where knowledge work gets done. The usage data may be more significant than the product news itself — if only 8.7% of Cowork sessions are coding tasks, the 'AI coding tool' framing massively undersells what's already happening. Businesses are deploying these agents for core administrative and operational workflows, and the race between OpenAI, Google, and Anthropic to capture that surface is accelerating. For enterprises evaluating AI strategy, the question is shifting from 'which chatbot' to 'which agent platform will run my background work.'