Claude Can Now Work Your Desktop on Team and Enterprise Plans

Human Verified

Share to: , and

A computer monitor with a glowing cursor on a warm-lit desk

Anthropic has brought computer use to Claude’s Team and Enterprise plans in beta, letting the assistant see a user’s screen and work in the desktop apps they allow. The feature is off by default, switched on only by an organisation’s owner, and is rolling out gradually.

What It Does

Computer use lets Claude operate a computer as a person would, clicking, typing and moving between applications. In the new Claude Desktop experience, it works in the desktop apps a user permits, and it also runs from Claude on the web and on phones where computer access is turned on.

The capability is not new: computer use existed as a research preview for developers, and a version for Pro, Max and Ultra users is live. What changes now is who can deploy it — businesses, with admin controls deciding which apps Claude may touch and who may switch it on.

Anthropic points administrators to a setup guide for turning the feature on, and the rollout is gradual, so not every organisation has it yet. No pricing change was announced, and the company has not said which desktop apps are supported beyond the ones each user allows.

Why Now

The rollout lands in a week when what an AI agent may see and touch on a user’s machine is the industry’s debate. Signal moved to block agents from reading its chats, and Meta’s Muse was found reading data through accessibility permissions. Anthropic’s answer is admin control, not open access.

The enterprise framing is the opposite of the consumer agent land grab. Consumer agents ask users to hand over their screens; Claude’s enterprise rollout asks an organisation’s owner to opt in, and keeps the switch off until they do. That is slower, but it is the path customers can approve.

Anthropic’s release notes underline the caution: the feature is off by default, and only an owner or primary owner can switch it on. The gradual rollout means some organisations will wait weeks, and the company has said nothing about when, or whether, the beta reaches smaller plans.

Our Read

Computer use is Anthropic’s agent capability, and putting it on business plans is when it stops being a demo and becomes a product companies must govern. The entry understates the shift: once Claude can operate desktop apps, it stops answering questions and starts doing work inside tools a company runs.

The cautious rollout is the right call, and it also tests Anthropic’s enterprise pitch. Admins will want to know which apps Claude can operate, what it can see, and what happens when it makes a mistake. The setup guide answers some of that, but real deployments will surface the rest.

Why It Matters

Agents that can see screens and operate apps are the next step beyond chat, and Anthropic is now selling that step to businesses. The move follows Claude’s arrival in Google Meet last week, and it puts Anthropic alongside Google, whose Gemini took over the Mac days earlier.

Background

Computer use first appeared as a developer preview in 2024, and Anthropic has spent the year building enterprise controls around it, from security centres to spend tracking. The Team and Enterprise beta, announced in release notes on November 20, offers the capability as a business feature for the first time.

Source: Anthropic’s release notes

Share to: , and

Faheem R. Khattak

Senior automation engineer and AI marketing journo covering LLMs, algorithms, analytics, core updates, Bing search, and technical SEO.

0 Comments

Leave a Reply

Optional Take Tag:

🛡️ Spam protection: Honeypot + Time-Trap Active

This site uses Akismet to reduce spam. Learn how your comment data is processed.