Anthropic adds computer control to Claude as Agentic AI heats up

Anthropic has expanded its AI assistant Claude with new computer use capabilities, allowing it to carry out tasks directly on a user’s device.

The update signals a broader shift among AI firms towards building autonomous agents that can complete real-world actions beyond chat.

Announced on Monday, the feature enables users to assign tasks remotely, including from a phone, with Claude executing them on a connected computer.

The move comes as competition heats up following the rapid rise of OpenClaw, which has drawn industry attention for similar agent-style functionality.

Claude computer control and task automation

With the latest update, Claude can open applications, browse the internet, and interact with files such as spreadsheets.

Anthropic demonstrated the feature in a video shared on Monday, where a user running late for a meeting asks Claude to export a pitch deck as a PDF and attach it to a calendar invite.

The AI then completes the task without manual input.

Users can initiate tasks by sending instructions from a phone, after which Claude executes them on their computer.

This expands its role from a conversational assistant to a tool capable of handling workflows across devices.

Anthropic said the feature is designed to integrate into everyday productivity tasks, enabling users to automate actions that would otherwise require direct interaction with a computer interface.

AI agent race driven by OpenClaw momentum

The rollout highlights a growing focus on so-called agentic AI, where systems operate independently to complete tasks.

Interest in this category surged after the release of OpenClaw, which quickly gained traction for its ability to connect with models from OpenAI and Anthropic.

OpenClaw allows users to send instructions through platforms such as WhatsApp or Telegram.

It runs locally on devices, giving it access to files and system functions, similar to Claude’s new capabilities.

The development has attracted attention from major players.

Nvidia CEO Jensen Huang said last week that OpenClaw represents the next major evolution after ChatGPT.

The company also introduced NemoClaw, an enterprise-focused version aimed at businesses.

OpenAI has also moved to strengthen its position in this space.

Last month, it hired OpenClaw creator Peter Steinberger as part of efforts to accelerate work on personal AI agents.

Safeguards and early stage limitations

Anthropic said the computer use feature is still in an early phase compared to Claude’s existing strengths in coding and text-based tasks.

The company noted that the system can make errors and that risks remain as threats evolve.

To address this, Anthropic said it has introduced safeguards designed to reduce misuse.

Claude is required to request user permission before accessing new applications, adding a layer of control over its actions.

The company also emphasised that ongoing improvements will be needed as the technology develops, particularly as more complex tasks are introduced.