Anthropic has released a major update that transforms Claude from a conversational assistant into something far more powerful: a remote digital operator capable of controlling your Mac.
The new capability, now available as a research preview for paying subscribers, represents a significant escalation in the race to build AI agents that can actually perform meaningful work on behalf of users.
From Chat to Action
Previously, interacting with Claude meant having conversations — useful for brainstorming, writing assistance, and information retrieval. But the new update fundamentally changes this dynamic by enabling Claude to take actions in the real world through your computer.
The feature is being rolled out across both Claude Cowork, Anthropic agentic productivity tool, and Claude Code, the company developer-focused command-line agent. Users can now assign tasks to Claude from anywhere — including via mobile phone through the newly expanded Dispatch feature — and return to find completed work.
A Complete Workflow Pipeline
Perhaps most significantly, Anthropic is extending Dispatch into Claude Code for the first time. This creates an end-to-end pipeline where users can issue instructions from anywhere — a mobile phone, while commuting, during a meeting — and return to find their requested deliverable completed.
Imagine sending Claude a message like Create a presentation deck for tomorrow investor meeting based on the Q4 report in my documents folder and having it automatically created, formatted, and saved ready for your review. This is the workflow Anthropic is now enabling.
Competition Heats Up
The release escalates the ongoing competition between AI labs to build the most capable agent systems. Microsoft recently announced Copilot Cowork integration with Claude as part of their Frontier Program, while Google Gemini and OpenAI GPT models continue advancing their agent capabilities.
What sets Anthropic approach apart is the depth of integration and the emphasis on providing a seamless mobile-to-desktop workflow. The ability to assign tasks via phone and have them completed on your workstation addresses a key pain point in current AI assistant paradigms.
Security and Control
Anthropic has implemented several safeguards for the new capabilities. The Mac control features are explicitly opt-in, and users maintain control over what applications Claude can access. Actions are logged and can be reviewed before execution if desired.
As AI systems become capable of taking real-world actions, these safety considerations become increasingly important. Anthropic thoughtful approach to gradual deployment and user control reflects broader industry awareness of the need for responsible AI development.
The Future of Work
The implications for knowledge work are significant. Tasks that previously required sitting at a computer — creating documents, writing code, managing files, drafting communications — can now be delegated to AI agents that work autonomously in the background.
This shift represents a fundamental change in how we think about human-AI collaboration. Rather than AI as a tool we actively operate, we are moving toward AI as autonomous colleagues capable of handling entire workflows with minimal supervision.
The research preview is now available for Claude subscribers. As the technology matures and more users gain access, we will likely see rapid evolution in how organizations integrate these capabilities into their daily operations.