Microsoft has introduced a new feature called "Computer Usage" for Copilot Studio, enabling AI agents to interact with websites and desktop applications. This functionality is similar to OpenAI's Operator and Claude's equivalent feature, aiming to help businesses build AI agents capable of automating task handling.

According to Charles Lamanna, Microsoft's Corporate Vice President of Business Applications and Industry Copilot, the "Computer Usage" feature allows AI agents to interact with websites and desktop applications by clicking buttons, selecting menus, and entering text on the screen. This capability enables AI agents to perform tasks without requiring direct API connections, meaning that if a human user can operate the application, so can the AI agent.

This new feature in Copilot Studio will support the creation of AI agents that automate data entry, conduct market research, and even process invoices. Microsoft states that the tool can detect changes in buttons and screens within applications or websites, ensuring continued smooth operation even after such modifications, without any interruptions or malfunctions.

Previously, Microsoft added a similar feature to the consumer version of Copilot, known as "Actions." This feature allows users to execute tasks like booking restaurants, purchasing event tickets, and shopping online while performing other activities in the background. However, the Actions feature is currently limited to use with specific partners. In comparison, Copilot Studio’s "Computer Usage" feature appears to offer greater flexibility in terms of accessible websites and applications.