Build from anywhere with Cursor for iOS

Summary

Cursor has released a native iOS application in public beta that lets developers start and oversee AI coding agents from a mobile device. The app supports two operating modes: launching persistent agents that run in isolated cloud virtual machines, or remotely directing agents that execute on a developer’s local machine. Both approaches allow tasks to be initiated outside the desktop environment, with progress tracked through lock-screen Live Activities and push notifications that signal completion, required input, or readiness for review.

Users select a repository and launch an agent through the same interface as the desktop client. They can choose from frontier models, dictate instructions via voice input, and apply slash commands to steer behavior. For local agents, a Remote Control feature maintains connectivity, with an optional setting that prevents the host machine from sleeping. Cloud agents operate asynchronously, generating not only code changes but also screenshots, logs, and runnable demos that can be inspected directly in the app before diffs are reviewed or pull requests are merged.

Work can be handed off between environments: a plan created locally can be transferred to the cloud for extended execution, while a cloud session can be moved back to the desktop for final local verification. The company notes that teams already combine these agents with external tools via MCP integrations to query observability data or summarize activity in other platforms. The beta is available to paid subscribers, with additional repository-agnostic chat capabilities planned to reduce the need for full codebase context on every task.

Why it matters

This update fundamentally shifts how product teams and builders interact with AI coding assistants, enabling asynchronous, mobile-first agent management. Dutch AI practitioners and SMEs can leverage this to improve incident response times and maintain development momentum outside traditional working hours.