Anthropic’s Claude Now Controls Your Computer

Anthropic’s Claude Now Controls Your Computer

Artificial intelligence is getting much more capable. Anthropic, a leading AI company, has recently released new features for its AI assistant, Claude. These updates allow Claude to interact with and control your computer, much like a human would. This significant development brings Claude closer to the capabilities previously seen in popular open-source AI projects like OpenClaw.

Claude’s New Computer Skills

Anthropic has been rapidly adding new abilities to Claude. The latest additions focus on enabling Claude to perform tasks directly on your computer. This means Claude can now open applications, browse the web, fill out spreadsheets, and manage files, all without direct human command for each step. This functionality is currently available in Claude’s Co-work and Code features, but it’s limited to Mac operating systems for now. The company is calling this new capability ‘computer use’.

These features are part of a larger push by Anthropic to make Claude a more active assistant. Previously, AI assistants primarily responded to text prompts. Now, they can take initiative and perform actions within your digital environment. This is a big step towards AI agents that can truly work alongside us, handling complex, multi-step tasks.

The Rise of AI Agents

The concept of AI agents that can perform tasks autonomously has gained huge traction recently. OpenClaw, an open-source AI agent, became incredibly popular because it could run on local hardware and manage files, emails, and calendars. It even learned about the user over time, becoming a personalized assistant. The demand for OpenClaw was so high that it reportedly caused shortages of Mac Mini computers.

However, OpenClaw required technical setup and had potential security concerns, as it ran on a user’s own machine. Anthropic’s new Claude features aim to offer similar power but with greater ease of use and built-in security. By hosting Claude on their servers, Anthropic handles the complex installation and security aspects, making it more accessible for everyday users.

Comparing Claude and OpenClaw

Many experts believe Anthropic’s new features offer about 90% of the functionality that users wanted from tools like OpenClaw. While OpenClaw offered deep control and data privacy by running locally, Claude’s approach is more user-friendly and less prone to security risks for the average person. Claude’s computer control is sandboxed, meaning it has limited access to your system, making it safer than a fully self-hosted agent that could potentially cause more damage if compromised.

For instance, Anthropic’s system requires users to grant specific permissions for Claude to access apps or perform certain actions. This built-in safety net is a significant advantage for consumers who may not be highly technical. Unlike OpenClaw, which can be complex to set up and manage, Claude’s integration is designed to be more straightforward, especially for those already using the Claude desktop app.

Dispatch: Mobile Control for AI Tasks

Beyond desktop control, Anthropic also introduced ‘Dispatch.’ This feature allows users to assign tasks to Claude from their mobile phones. Imagine telling Claude to scan your emails every morning or pull a report every Friday via a simple app on your phone. Claude then handles these tasks in the background on your computer. This mobile integration mirrors the convenience that messaging apps offered for controlling OpenClaw.

While a dedicated Claude app is currently available, there are signs that Anthropic is working on a more general mobile automation framework, internally codenamed ‘orbits.’ This suggests a future where users can interact with Claude for complex tasks through various mobile apps and services.

Why This Matters

The rapid advancement of AI assistants like Claude marks a significant shift in how we interact with technology. AI is moving from simple question-answering to active task completion. This means AI can now handle routine, time-consuming jobs, freeing up human users for more creative and strategic work.

For businesses, this could mean automating customer service, data analysis, and content creation. For individuals, it could mean personalized AI assistants that manage schedules, organize information, and even help with creative projects. The increased accessibility and safety of Anthropic’s approach mean these powerful tools are likely to become widespread much faster than previous, more technical solutions.

Open Source vs. Commercial AI

While Anthropic’s offerings are powerful, open-source projects like OpenClaw still hold value. OpenClaw is free and open-source, allowing users to run it with any AI model they choose, offering flexibility and avoiding vendor lock-in. It also runs on multiple operating systems, including Linux and Windows, and can be deployed on low-cost hardware like Raspberry Pis, unlike the current Mac-only limitation of Claude’s computer control.

Furthermore, the open-source community continually develops new ‘skills’ and extensions, fostering a collaborative and adaptable ecosystem. For highly technical users or those with strict data privacy needs, OpenClaw and similar projects may remain the preferred choice. However, for the vast majority of users, Anthropic’s polished, secure, and easy-to-use AI agent represents a major leap forward.

Anthropic’s rapid release cycle and focus on user-friendly features suggest a strong commitment to bringing advanced AI capabilities to a broad audience. The company is shipping new features at an impressive pace, with many improvements rolled out in just a few weeks. This aggressive development strategy indicates that AI assistants are set to become even more integrated into our daily lives.

Source: Claude just killed OpenClaw (YouTube)