Anthropic has announced that users can now let Claude take control of their computer to perform various tasks. In order to make this feature work, they just need to be on a qualifying subscription plan. The latest update makes Claude Code and Claude Cowork tools capable of opening files, using the browser as well as running multiple tools.
Introduced in January, Claude Cowork is an iteration of the Claude Code AI agent for programmers. It is said to have been designed for more casual users.
How It Works?
Once enabled, the Claude AI chatbot prioritises connectors to supported services like Google workplace suite or Slack, according to Anthropic. However, if the tool or connector remains unavailable, even then it will be able to execute the assigned task. Claude manually performs the task by typing or moving the cursor, as if it is using the keyboard and mouse.
Before taking these actions, Claude should ask for permission. Anthropic has recommended not to use this feature in order to handle sensitive information as a precaution.
When performing the assigned tasks, Claude uses your computer by scrolling and clicking around. However, the key difference is that Claude it will seek permission beforehand. This means, users will be able to stop Claude from performing a task at any time.
"It opens your apps, navigates your browser, fills in spreadsheets — anything you'd do sitting at your desk," read a post from Claude on X.
Initially, the Claude computer remains available to Claude Pro and Claude Max subscribers on macOS. As of now, it is still in a research preview and is expected to be adjusted based on Anthropic's user feedback.
Moreover, it will support use with Anthropic's Dispatch feature that allows users to message the chatbot in a single continuous conversation across phone and desktop.
This comes in the wake of the viral explosion of OpenClaw framework earlier this year. Recently, Nvidia debuted NemoClaw, its own framework for easily setting up and installing OpenClaw, with a few security settings. While giving full access to the chatbot can be convenient for certain tasks, it does leave the computer vulnerable to attacks.
Anthropic said it has implemented safeguards to further minimise risks like prompt injections. According to the company, the system automatically scans for this.
Also Read: Apple Eyes AI Resurgence As WWDC Kicks Off On June 8
Since the feature is new and might contain errors, the company has suggests not to use apps that handle sensitive data.
Essential Business Intelligence, Continuous LIVE TV, Sharp Market Insights, Practical Personal Finance Advice and Latest Stories — On NDTV Profit.