Anthropic launched Cowork this week, a desktop agent that gives its Claude AI direct access to computer folders for autonomous file management. The feature transforms Claude from a conversational chatbot into an active digital teammate capable of reading, editing, and creating files without user intervention.
Cowork operates through the Claude desktop app on macOS, requiring users to grant folder-level permissions. Once access is established, Claude can execute multi-step tasks like organizing downloads, generating expense reports from receipt images, and drafting documents from scattered notes. The system works with common file formats including PDFs, spreadsheets, and text documents.
The tool represents a strategic expansion of Anthropic's Claude Code technology beyond developers. Built on the same agentic framework as Claude Code, Cowork repackages coding capabilities for general knowledge workers. Anthropic developed the feature after observing users employing Claude Code for non-programming tasks like tax preparation, receipt management, and file organization.
Access remains limited to Claude Max subscribers, Anthropic's power-user tier priced between $100 and $200 per month depending on usage, with the feature currently in research preview on macOS. Users on other plans can join a waitlist for future availability. The company describes the experience as "much less like a back-and-forth and much more like leaving messages for a co-worker."
Cowork integrates with Claude's existing connector ecosystem, enabling connections to external services like Google Drive and Slack. When paired with the Claude in Chrome extension, the agent can complete browser-based tasks. The system plans tasks internally before execution, showing progress updates as it works through queued assignments.
Security concerns accompany the expanded access. Anthropic warns about prompt injection attacks where malicious instructions hidden in external content could compromise the agent. The company also cautions that vague prompts might lead to unintended file deletions, recommending users avoid sensitive data during the preview phase.
The launch follows Anthropic's healthcare announcement earlier this week, positioning the company against OpenAI's ChatGPT health tools. Cowork development reportedly took less than two weeks using Claude Code itself, according to Anthropic's head of Claude Code Boris Cherny.
Industry analysts view Cowork as part of a broader shift toward agentic AI systems that perform actions rather than just provide information. The move could pressure Microsoft and Google to accelerate their own desktop-native agent development. Market observers suggest the next 6-12 months will determine which company captures dominant market share in this emerging category, as Apple continues to develop high-end desktop hardware like the rumored M5 Max iMac Pro.
Anthropic emphasizes that Cowork remains a research preview with active safeguards under development. The company advises precise instructions and recommends starting with non-critical tasks like renaming PDFs or organizing downloads before advancing to more complex workflows.
