- Who it helps
- Developers and desktop users evaluating automation across graphical applications.
- Availability
- GitHub announced a public preview for macOS and Windows on October 1 in Copilot CLI and the Copilot app. Organization policy can disable it. Current plan eligibility and per-task cost were not fully verified.
What the feature can attempt
The announcement describes clicking, typing, scrolling and moving information through applications. It includes an expense-report example. This establishes intended capabilities, with task-specific reliability still requiring a test.
Approval and operating-system access
Copilot requests approval before controlling an application. Users can review or reset previously allowed apps. macOS also requires Accessibility and Screen Recording permissions.
What you can do
- Read the preview’s current permission and access documentation before enabling control of an application. Open reference ↗
Where the evidence stops
No permission was granted and no application workflow was run. The preview announcement does not establish reliable completion of arbitrary desktop tasks.
Sources read. Sources checked October 21, 2026. No product workflow or benchmark was executed.
Sources & further reading
Web · official
Copilot computer-use public preview ↗
Establishes supported desktop platforms and the October public-preview announcement.
GitHub · page checked 2026-10-21 · Published 2026-10-01
Web · official
Copilot computer-use concepts ↗
Explains application control and permission requirements.
GitHub · page checked 2026-10-21
Coverage note: Independent hands-on testing and cross-platform discussion are incomplete. No missing links, creator identities or engagement counts are inferred.
The record over time (2)
Dated source checks and published explanations. An observation date can differ from the original release date.
- · Source recheck
Copilot computer use lets an agent read and operate desktop application interfaces, including software without a usable API.
- · Added to Current
Copilot computer use lets an agent read and operate desktop application interfaces, including software without a usable API.