On January 12, 2026, Anthropic introduced a groundbreaking new feature called Claude Cowork. This innovative addition is constructed upon the foundational architecture of Claude Code. Its primary objective is to broaden the reach of the AI model Claude's agent capabilities, making them accessible to individuals who are not professional developers. By doing so, it facilitates a more seamless and efficient interaction between humans and machines.

With Claude Cowork, users have the ability to grant Claude permission to access particular folders. Once authorized, Claude can perform a range of actions, including reading, editing, or even creating files. Additionally, it offers support for task queues and parallel processing, further enhancing its functionality.

At present, this feature is being offered as a 'research preview.' Its availability is restricted to subscribers of Claude Max who are using the macOS application. The monthly subscription fees for this service range from $100 to $200.

Anthropic also takes this opportunity to remind users about potential risks associated with using Claude Cowork. These risks include the possibility of unintended file operations that may occur if instructions are not clear enough. Moreover, there are security vulnerabilities to be aware of, such as prompt injection attacks.