Anthropic has launched Claude Co-work, an evolution of Claude Code, expanding its AI assistant from coding tasks to a wide range of general productivity activities for non-developers, with features like a secure virtual machine, browser automation, and cloud integration. Currently available for macOS with a Claude Max subscription, Claude Co-work emphasizes user safety, privacy controls, and intuitive workflows, with plans for broader access and ongoing improvements.
Anthropic has just launched Claude Co-work, an evolution of their previous product, Claude Code. While Claude Code was designed specifically for programmers, allowing them to interact with their codebase and automate coding tasks, users quickly discovered it could be used for a wide range of non-coding activities. These included tasks like vacation research, building slide decks, cleaning up emails, canceling subscriptions, recovering files, and even controlling smart devices. Recognizing this broader utility, Anthropic developed Claude Co-work to make these capabilities more accessible to non-developers and general knowledge workers.
Claude Co-work is designed to be user-friendly and includes several novel features aimed at safety and ease of use. One of the key features is a built-in virtual machine (VM) for isolation, which ensures that the tool operates in a secure environment, protecting users’ computers from unintended changes or security risks. This is particularly important for less technical users who may be wary of giving an AI agent access to their files and system. The product also offers out-of-the-box support for browser automation and integrates with various cloud AI data connectors, allowing it to interact with services like Google Drive and Box.
The demo showcased how Claude Co-work can perform practical tasks, such as organizing a cluttered desktop. The AI reads the contents of the desktop, asks clarifying questions, and then organizes files into well-categorized folders almost instantaneously. The workflow is similar to Claude Code: the AI creates a plan, executes it step by step, and verifies each action. This process is designed to be intuitive and familiar for those who have used Claude Code before, but is now tailored for a broader range of everyday tasks.
Currently, Claude Co-work is only available for macOS and requires a Claude Max subscription, so its availability is somewhat limited at launch. However, the developers emphasize that this is just the beginning, and the product will continue to improve and become more widely accessible over time. Users can queue up multiple instructions, allowing for efficient multitasking, and the system only accesses specific folders that the user explicitly grants permission to, providing fine-grained control over privacy and security.
Safety remains a top priority, with Claude Co-work asking for user confirmation before taking significant actions, such as deleting files or making major changes. The developers also warn about the risk of prompt injections—malicious attempts to alter the AI’s behavior through manipulated content encountered online. While Anthropic has implemented sophisticated defenses, they acknowledge that no system is completely immune. Early feedback from users has been very positive, with many praising its capabilities, and further updates and improvements are expected as more people begin to use and test the product.