Claude on Windows 2026: How to Enable & Full Feature List [Step-by-Step]

Anthropic just announced that computer use in Claude Cowork and Claude Code Desktop is now available on Windows. The feature, which launched on macOS on March 23, 2026, has been one of the most anticipated expansions for the 70% of desktop users who run Windows. As of today, Windows users on Pro and Max plans can enable Claude to directly control their computer — clicking, typing, scrolling, and navigating apps on their behalf.

The announcement came via Anthropic's official X account: "Computer use in Claude Cowork and Claude Code Desktop is now available on Windows. You can now enable Claude to use your computer to complete tasks. It opens your apps, navigates your browser, fills in spreadsheets — anything you'd do sitting at your desk."

What computer use actually does

Computer use gives Claude a new fallback layer when completing tasks. The existing tool hierarchy in Cowork works like this: Claude first reaches for connectors — direct integrations with services like Gmail, Google Drive, Slack, or Google Calendar. These are fast and reliable. When a connector exists, Claude uses it. When one does not, Claude can now fall back to controlling your screen directly — navigating apps, clicking buttons, typing into fields, and scrolling through content exactly as a human would.

This matters because the long tail of apps that will never have a dedicated Claude connector is enormous. Proprietary internal tools, niche software, desktop-only applications, legacy systems — none of these have connectors, and none of them will. Computer use bridges that gap by letting Claude interact with any app that runs on your screen, regardless of whether Anthropic has built a direct integration for it.

Specific things Claude can do with computer use enabled: open files in any desktop application, navigate the Chrome browser to complete web-based tasks, fill in spreadsheets and forms, run developer tools and terminal commands in Claude Code, interact with File Explorer, and execute multi-step workflows that span multiple applications in sequence.

How it works with Dispatch

Computer use pairs directly with Dispatch — Anthropic's feature that lets you assign Claude tasks from your phone and have them executed on your desktop while you are away. The combination is where the capability becomes genuinely useful for daily workflows. You send a task from your phone on the train. Your Windows desktop wakes up, Claude opens the relevant apps, completes the work, and sends you a push notification when it is done or needs approval. You come back to finished work.

Practical examples of this pairing: creating a morning briefing by pulling from multiple apps while you are commuting; making changes in your IDE, running tests, and opening a pull request while you are in a meeting; or keeping a long-running project moving according to an initial plan you set the night before. Dispatch-spawned Claude Code sessions can use computer use too, with one difference — app approvals expire after 30 minutes and re-prompt, rather than lasting the full session.

How Claude prioritizes what to do

Anthropic built a clear hierarchy into how Claude handles tasks with computer use available. Claude always reaches for the most precise and reliable tool first. The order is: connectors first, browser second, screen interaction last. If a task can be done through a direct Gmail integration, Claude uses that — it is faster, more reliable, and less error-prone than navigating the Gmail interface on screen. Browser control comes next for web-based tasks without a connector. Screen interaction is the last resort, used only when neither of the first two options is available.

This connector-first philosophy is deliberate. Screen navigation is inherently slower and more fragile than a direct API integration. Claude's help documentation is explicit about this: "Screen interaction is slower than connectors. When Claude works through your screen instead of a direct integration, tasks take longer. Where possible, connect the tools you use most."

Safety controls built in

Computer use runs outside the virtual machine that Cowork normally uses for file operations and command execution. This means Claude is interacting with your actual desktop rather than an isolated sandbox — which makes the safety controls more important, not less.

Anthropic has built several safeguards into the feature. Per-app permissions: Claude asks before accessing each application for the first time. Some sensitive categories are blocked by default — investment and trading platforms and cryptocurrency apps cannot be accessed by computer use without explicit override. App blocklist: users can add any application to a blocklist and any Claude request to use a blocked app will be automatically denied. Action review: the system scans for signs of prompt injection when Claude uses your computer — a specific attack vector where malicious content on screen tries to redirect Claude's actions. Users can stop Claude at any point during a task.

Anthropic's guidance on sensitive data is unambiguous: do not use computer use for regulated workloads, financial credentials, health information, or passwords at this stage of the research preview. This is explicitly early-stage technology and the safeguards, while real, are not yet comprehensive enough for high-stakes environments.

How to enable it on Windows

Make sure you have the latest version of Claude Desktop installed — download or update at claude.ai/download. Open the desktop app and go to Settings → General under Desktop app. Toggle on Browser use first, then toggle on Computer use. When toggling Computer use on, you will need to accept the terms by clicking Turn on. Grant Accessibility and Screen recording permissions when prompted — these are required for Claude to see and interact with your screen. Then open Cowork or Claude Code in the desktop app and start a session.

One hardware note from the official Claude Code documentation: on Windows, Cowork is available on all supported hardware — there is no Apple Silicon restriction equivalent. For the Code tab specifically, Git is required on Windows. Download Git for Windows, install it, and restart the app before using the Code tab.

Who gets it and what it costs

Computer use is available to Claude Pro ($20/month) and Claude Max ($100–$200/month) subscribers. Team and Enterprise plans do not have access to computer use at this time. Free plan users cannot access the feature. The feature is a research preview — Anthropic is gathering feedback on where it works well and where it falls short, consistent with how they rolled out Cowork itself earlier this year.

Working with computer use consumes more usage allocation than standard chat. Complex multi-step tasks that involve screen navigation require more compute resources. Pro plan users will hit usage limits faster on computer use tasks than on regular conversations. Max plan subscribers get significantly higher limits. If you hit limits frequently, batching related work into single sessions and using standard chat for simpler tasks that do not need screen control is the practical workaround.

What this means for Windows users

The Windows expansion of computer use is a meaningful milestone. When Cowork launched in January 2026 it was macOS-only. When computer use launched in March 2026 it was again macOS-only. The pattern of macOS-first, Windows-to-follow has now completed its second cycle, and Windows users have full access to the same autonomous desktop AI that Mac users have been testing for the past week.

For the roughly 70% of desktop users who run Windows, this is the first opportunity to try Claude's most autonomous capability in a real workflow. The honest caveat from Anthropic's own documentation remains: this is a research preview, complex tasks sometimes need a second try, and screen interaction is slower than direct integrations. Start with low-stakes, read-only tasks. Watch the first run. Expand from there as you build confidence in how Claude interacts with your specific apps and workflows.