Microsoft launches Copilot Tasks, AI agent that automates work across apps and browser

Microsoft announced Copilot Tasks on Thursday, a new AI-powered automation feature designed to handle repetitive work in the background using cloud-based infrastructure.

The feature operates as an agent that can execute tasks across multiple applications and browsers without consuming local device resources. Users describe what they need using natural language, and Copilot Tasks handles execution. Tasks can be configured to run once, on a schedule, or on a recurring basis. The system provides a completion report once work is finished.

According to Microsoft, the feature supports a range of use cases including scheduling appointments, generating study plans, and managing other administrative work. The exact scope of supported applications and integrations was not detailed in the announcement.

Copilot Tasks is currently in preview status. Microsoft has not disclosed pricing, general availability timeline, or which specific Copilot plans will include the feature. The company also did not specify which underlying AI models power the automation or provide technical details about how the agent handles complex multi-step workflows.

The release fits Microsoft's broader strategy of integrating AI agents into productivity tools. Similar automation features have been announced by competitors including OpenAI with its Agent Builder and Google with tools in Workspace.

What this means

Copilot Tasks represents a shift from AI assistants that answer questions to AI systems that take autonomous action across multiple services. The cloud-based execution model addresses a practical constraint: running complex multi-app workflows on user devices would drain resources. Success depends on whether the agent can reliably execute tasks without constant supervision and whether Microsoft can integrate deeply enough with third-party applications to be genuinely useful beyond Microsoft's own ecosystem. The lack of disclosed pricing and availability timeline suggests the feature is still early in development.