Anthropic pushes Claude Cowork

Anthropic has moved Claude Cowork out of preview and into general availability on paid plans, adding the admin controls companies usually demand first. (support.claude.com) Anthropic said on April 9 that Claude Cowork is now generally available on macOS and Windows through the Claude Desktop app. The same release added usage analytics, Analytics Application Programming Interface support, OpenTelemetry monitoring, role-based access controls, and group-based setup through System for Cross-domain Identity Management. (support.claude.com) Claude Cowork is Anthropic’s desktop agent for “multi-step knowledge work,” meaning a user gives it a goal and it works across local files, folders, and apps to produce a finished document or other deliverable. Anthropic says it built the product after non-technical teams inside the company started using Claude Code for jobs like data mining and tool-building that went beyond chat. (anthropic.com) The new controls are aimed at company-wide rollouts, not one-off experiments by a few technical users. Anthropic’s own rollout webinar for April 16 says administrators can now assign custom roles by group, cap spending by team, monitor usage in dashboards and through an analytics feed, and set granular permissions for connectors and tools. (anthropic.com) That makes Claude Cowork look less like a personal assistant and more like managed workplace software. Anthropic is packaging one desktop deployment for two audiences at once: Cowork for knowledge workers and Claude Code for engineering teams. (anthropic.com) The timing fits a broader race among artificial intelligence companies to sell agents inside large organizations, where security reviews and budget controls often slow adoption more than model quality does. SiliconANGLE reported April 9 that Anthropic framed the update around “organization-wide controls” for corporate deployments, while OpenAI made its own enterprise push the same day with lower pricing around Codex tools. (siliconangle.com) Anthropic is also emphasizing connector governance, which is the question every information technology team asks when an agent can reach internal systems. In the April 16 event description, the company says admins can scope access to Cowork, Claude Code, web search, connectors, and other capabilities by group rather than turning everything on for everyone. (anthropic.com) The company is pairing that pitch with compliance language meant for procurement teams. Anthropic’s trust center says Claude for Enterprise has certifications or attestations including SOC 2 Type 2, International Organization for Standardization 27001, International Organization for Standardization 42001, Health Insurance Portability and Accountability Act readiness, and National Institute of Standards and Technology 800-171. (trust.anthropic.com) Anthropic is still describing Cowork as a system that works under human oversight, not a tool that makes final decisions on its own. The closer now is practical, not philosophical: Claude Cowork can be bought on paid plans, deployed on Mac and Windows, and governed with the same kinds of controls companies use for other shared software. (anthropic.com)