Anthropic has taken the most significant step yet toward making AI agents a practical reality for everyday users. The company announced that Claude can now directly control Mac computers鈥攃licking buttons, opening applications, typing into fields, and navigating software on the user’s behalf. It’s available immediately as a research preview for paying subscribers.
From Conversational AI to Digital Operator
The computer use feature transforms Claude from a chatbot that answers questions into something closer to a remote digital operator. When a user assigns a task, Claude can now interact with applications the way a human would: clicking, typing, scrolling, and navigating screens.
The feature works through a layered priority system. Claude first checks whether direct connectors exist for services like Gmail, Google Drive, Slack, or Google Calendar鈥攖hese provide the fastest and most reliable execution path. If no connector is available, Claude falls back to navigating Chrome via Anthropic’s browser extension. Only as a last resort does it interact directly with the desktop screen.
This hierarchy reveals how Anthropic is thinking about reliability versus reach. As the company’s documentation explains: “Pulling messages through your Slack connection takes seconds, but navigating Slack through your screen takes much longer and is more error-prone.” Screen-level interaction is the most flexible mode but also the slowest and most fragile.
Dispatch: Your iPhone as an AI Remote Control
The strategic masterstroke may be the pairing of computer use with Dispatch, which launched last week for Claude Cowork and is now extending to Claude Code. Dispatch creates a persistent conversation between Claude on your phone and Claude on your desktop.
Users pair their mobile device with their Mac by scanning a QR code. From that point, they can text Claude instructions from anywhere, and Claude executes those instructions on the desktop鈥攚hich must remain awake and running the Claude app鈥攖hen sends back results.
Use cases range from mundane to ambitious: having Claude check email every morning, pull weekly metrics into a report template, organize a cluttered Downloads folder, or compile a competitive analysis from local files into a formatted document. Scheduled tasks allow users to set a cadence once鈥?every Friday,” “every morning”鈥攁nd let Claude handle the rest without further prompting.
Agentic AI as Infrastructure
The combination of Dispatch and computer use represents a paradigm shift. As one early user on social media noted: “Combine this with /schedule that just dropped and you’ve basically got a background worker that can interact with any app on a cron job. That’s not an AI assistant anymore, that’s infrastructure.”
The stakes of this competition are no longer theoretical. Reuters reported that OpenAI is actively courting private equity firms in what it described as an “enterprise turf war with Anthropic.” The ability to ship working agents is becoming the decisive weapon in that battle.
Current Limitations and Research Preview Status
Anthropic is calling this a research preview for a reason. Early hands-on testing suggests the feature works well for information retrieval and summarization but struggles with complex, multi-step workflows requiring interaction across multiple applications.
When Claude does interact with the screen, it takes screenshots to understand what it’s looking at鈥攚hich means Claude can see anything visible on the screen, including personal data. Anthropic has guardrails to prevent Claude from engaging in stock trading, inputting sensitive data, or gathering facial images, but the company acknowledges these “aren’t absolute.”
The feature is available to Claude Pro (/month) and Max (-/month) subscribers but currently only on macOS. Windows support is expected to follow.
The Future of Work
We’re witnessing the first concrete steps toward what many have called the “agentic AI” era鈥攁n shift from AI that advises to AI that acts. The implications for productivity are significant: tasks that previously required a person at a keyboard can now happen autonomously in the background.
Whether the reliability challenges can be overcome in the near term remains to be seen. But Anthropic’s bet is clear: the future of enterprise AI isn’t just about answering questions. It’s about getting work done.
Screenshots from Research and Coverage
