Anthropic has announced that its AI assistant Claude can now directly control a user's computer to complete tasks autonomously via mobile messages, marking a key advancement in AI agent development.
New Capability Launch
Anthropic's update enables Claude to perform actions on a user's device based on prompts sent from a phone. The AI can:
- Open applications
- Navigate web browsers
- Fill in spreadsheets
- Export files (e.g., PDFs) and attach them to emails or calendar invites
Demonstration and Use Case
In a video posted Monday, Anthropic demonstrated Claude handling a scenario where a user, running late for a meeting, asks Claude to export a pitch deck as a PDF and attach it to the meeting invite, which Claude completes without further intervention.
Context: The AI Agent Race
This development is part of a broader industry effort to create "agents" that operate autonomously. The viral OpenClaw tool, which connects to AI models from OpenAI and Anthropic and runs locally on devices, has accelerated this trend by allowing task completion through apps like WhatsApp.
Industry Response
Nvidia CEO Jensen Huang recently described OpenClaw as "definitely the next ChatGPT," highlighting competitive dynamics. Nvidia has also launched NemoClaw, an enterprise-grade version of OpenClaw, to address business applications.