Anthropic says Claude can now use your computer to finish tasks for you in AI agent push

Anthropic says Claude can now use your computer to finish tasks for you in AI agent push

Anthropic has unveiled an exciting new feature for its AI agent, Claude, which now has the ability to operate directly on a user's computer to complete various tasks. This announcement marks a significant step in the company’s ambition to develop AI agents capable of rivaling the popular OpenClaw. Users can now send tasks to Claude via their mobile devices, allowing the AI to execute actions on their computers. For instance, in a demonstration video, a user who is running late for a meeting instructs Claude to export a pitch deck as a PDF and attach it to a calendar invite. The video illustrates Claude successfully performing this task, showcasing its new functional capabilities. This latest development highlights a growing trend among AI companies striving to create autonomous agents that can manage tasks for users around the clock. The emergence of agentic abilities gained considerable attention following the viral success of OpenClaw, which integrates AI models from both OpenAI and Anthropic. Users can interact with OpenClaw through widely-used messaging apps like WhatsApp or Telegram, allowing it to perform tasks locally on their devices while accessing their files. Nvidia's CEO, Jensen Huang, recently referred to OpenClaw as 'definitely the next ChatGPT,' reflecting the competitive landscape as tech companies rush to introduce their versions of personal AI agents. Nvidia announced its own enterprise-grade version, NemoClaw, last week, further intensifying the race. In a strategic move, OpenAI recently hired Peter Steinberger, the creator of OpenClaw, as part of its efforts to advance the next generation of personal agents. However, Anthropic has issued a warning regarding Claude's computer capabilities, stating that they are still in the early stages compared to its proficiency in coding and text interaction. The company acknowledged that while Claude can make errors, it has implemented safeguards designed to mitigate risks and ensure user control, including requiring permission before accessing new applications. Additionally, users can utilize Dispatch, a recently launched feature in Claude Cowork, which enables ongoing conversations with Claude from either a mobile or desktop device, allowing for seamless task assignment.

Sources : CNBC

Published On : Mar 24, 2026, 10:15

Mobile
Kid-Safe Phone Innovations: A New Wave of Parental Control

With rising concerns among parents regarding unrestricted smartphone access for children, numerous companies are steppin...

TechCrunch | Jul 17, 2026, 16:25
Kid-Safe Phone Innovations: A New Wave of Parental Control
Gadgets
Transform Your Comfort: The Dyson Hot+Cool HF1 Review

The Dyson Hot+Cool HF1 is an innovative 2-in-1 appliance designed to provide optimal comfort throughout the year, seamle...

TechCrunch | Jul 17, 2026, 18:25
Transform Your Comfort: The Dyson Hot+Cool HF1 Review
Cybersecurity
Florida Man Arrested for Allegedly Using Malware-Infested Games to Steal Cryptocurrency

U.S. authorities have taken a significant step in combatting cybercrime with the arrest of a 21-year-old Florida residen...

TechCrunch | Jul 17, 2026, 16:45
Florida Man Arrested for Allegedly Using Malware-Infested Games to Steal Cryptocurrency
Startups
Final Call for Australian Startups: Seize Your Chance at Stripe x Startup Battlefield

Time is running out for aspiring Australian entrepreneurs! In just 48 hours, the opportunity to showcase your startup at...

TechCrunch | Jul 17, 2026, 23:25
Final Call for Australian Startups: Seize Your Chance at Stripe x Startup Battlefield
AI
Xi Jinping Advocates for Open-Source AI Collaboration at Global Conference

At the World Artificial Intelligence Conference held in Shanghai, Chinese President Xi Jinping emphasized the necessity ...

Business Insider | Jul 17, 2026, 17:10
Xi Jinping Advocates for Open-Source AI Collaboration at Global Conference
View All News