Anthropic has quietly integrated its Cowork agent capabilities into the main Claude application, marking a strategic divergence from the high-profile consumer launches of OpenAI’s Dots and Meta’s Muse. The update, which arrived two weeks ago, folds the functionality that previously allowed users to run scheduled jobs after closing their laptops directly into the core Claude experience for Pro and Max subscribers. This move bypasses the need for a standalone app or a flashy mascot, instead leveraging the existing user base of developers and enterprise teams already building with Claude.

The Always-On Agent Arms Race

The integration comes amidst a frenzied release cycle for autonomous agents. OpenAI launched Dots at DevDay on Tuesday, offering always-on agents with dedicated cloud computers and browsers, accessible via ChatGPT Pro plans ranging from $100 to $500 a month. Meta’s Muse, launched prior to Dots, has reportedly crushed mobile install numbers, with Apptopia estimating 359,000 daily U.S. users in its first 12 days compared to ChatGPT’s 231,000 at the same stage. xAI also entered the fray with Grok Bot in August, reaching 418,000 weekly users by mid-September. Anthropic’s approach is notably different: it shipped to paying subscribers rather than chasing viral consumer adoption, betting that recurring professional usage outweighs raw download metrics.

Copycat Features and Rapid Iteration

The timeline of these releases reveals how thin the ice is in the agent space. Anthropic’s Cowork features, including cloud scheduling, shipped on July 7, just two days before OpenAI’s Work mode. Meta’s Nat Friedman admitted that Muse was heavily inspired by OpenClaw, an open-source project that itself evolved from a weekend hack by Peter Steinberger. The SOUL.md personality file in Muse is reportedly nearly identical to OpenClaw’s. In a recent test by Jessica Wachtel, Claude and ChatGPT tied on accuracy for developer tasks, with ChatGPT being faster and Claude more thorough. This suggests that feature parity is rapidly being achieved, shifting the competitive advantage from capability to distribution and trust.

Security and Enterprise Friction

For enterprise buyers, the integration of autonomous agents raises significant security questions. An always-on agent holds credentials and acts autonomously, creating potential vectors for risk. xAI’s documentation explicitly warns users not to treat Grok Bots as a security boundary, while the OpenClaw Foundation notes that many IT departments ban such platforms outright. Anthropic leans into caution, with Claude asking for permission before acting unless configured otherwise, and maintaining audit trails via Claude Tag in Slack. Cat Wu noted that an internal version of these tools accounts for about 65% of product team code changes, indicating that for developers, the utility of automated QA bots and invoice checks outweighs the friction of manual supervision.

Key Takeaways

  • Anthropic integrated Cowork into Claude to target existing Pro and Max subscribers rather than chasing consumer downloads.
  • Meta’s Muse saw 359,000 daily U.S. users in its first 12 days, outpacing ChatGPT’s early growth metrics.
  • OpenAI’s Dots require a ChatGPT Pro or Enterprise plan, positioning them as premium products due to high compute costs.
  • Anthropic’s approach emphasizes security defaults and audit trails, contrasting with the more open, consumer-first models of rivals.

The Bottom Line

Anthropic is playing the long game by embedding agents into the workflows of people who already pay for AI, rather than burning cash to win the app store war. While Meta and OpenAI chase scale, Anthropic is banking on retention and trust.