Microsoft has made computer-using agents in Copilot Studio generally available, rolling the feature out to all commercial geographies as part of the platform’s May 2026 update. The announcement marks a meaningful shift in enterprise automation: agents that do not just answer questions but actively operate software on behalf of human workers.
What Computer-Using Agents Actually Do
A computer-using agent is given the same tools a human employee has: a browser, a screen, a keyboard, and the ability to read what is on the page and decide what to do next. It uses vision and reasoning to navigate live interfaces rather than hard-coded selectors that break the moment a vendor updates their user interface. That distinction matters enormously for British organisations running legacy enterprise software — ERP systems, government portals, and supply chain platforms — that have never exposed a clean API.
Traditional robotic process automation, which many UK organisations deployed over the past decade, locked onto specific screen coordinates or HTML elements. When the interface changed, the automation broke and required manual re-engineering. Computer-using agents adapt to interface changes in the same way a human would: by reading the screen and reasoning about the next step.
The Build 2026 Context
The GA release comes days ahead of Microsoft Build 2026, running 2–3 June, where the company is expected to announce a Windows Agent Runtime that positions the operating system itself as a native platform for autonomous agents. For UK IT teams, the Copilot Studio GA is the production-ready version they can deploy today.
Governance and Security Considerations
Microsoft has simultaneously made Agent 365 generally available at $15 per user per month, providing a unified control plane to govern and secure AI agents across Microsoft, AWS, and Google Cloud environments. For regulated UK industries — financial services, healthcare, and the public sector — the ability to inventory, audit, and block AI agents is not optional. The practical question for UK IT leaders is not whether to deploy AI agents but how to build the governance layer that keeps those agents accountable before a compliance incident forces the conversation.
