Cursor Brings AI Coding to Your Phone: What Mobile Agent Management Actually Changes

Cursor has released a public beta of its iOS app, now available to all paid subscribers. The app lets you access Cursor's AI coding environment on your phone or tablet, and more importantly, it lets you start and manage long-running coding tasks — called agents — from wherever you are.

The key feature here is agent orchestration on mobile. In Cursor, "agents" are automated processes that can work on code tasks without you being at your desk. They can refactor code, run tests, and iterate on projects over hours, running in the background. Until now, you had to be at your computer to start these tasks. The iOS app changes that. You can now kick off a multi-hour coding job from your phone, watch its progress in real time, and jump in to redirect it if needed — all without being at a workstation. The release was announced on 28 June 2026 in the Cursor changelog.

For developers already using Cursor's agents, the friction point was always the handoff: you start a task at your desk, but then you're stuck there waiting for it to finish, or you walk away and lose track. Mobile smooths that out. You can queue work from anywhere, watch the agent's output as it works, and step in to adjust course without needing to be at a computer. It is a narrow capability — you are not writing code on your phone — but it fits how longer-running AI tasks actually get used: in bursts of attention throughout your day, not in one long continuous session.

The beta is limited to paid tiers. This is Cursor's standard approach: roll out new features to paying subscribers first, gather feedback, then open it wider. Since most active Cursor users are on paid plans, the feature is essentially available to everyone who needs it, not stuck behind a waitlist.

Now, the fact that this is still beta matters. Running software on iOS comes with constraints that desktop environments do not have: the phone can shut down background tasks after a certain amount of time, network connections can drop, and apps have limited access to system resources. The real test will be how the iOS app handles agents that run longer than the phone's background limits, or what happens if your network flickers mid-task. These edge cases are exactly what beta testers are there to find. That stress test will determine whether this becomes a daily tool or just a convenient way to check in on things.

The direction is clear. As AI coding tools shift from simple autocomplete toward genuine multi-step autonomy, the interface becomes secondary to the ability to delegate work and then check back in later. A desktop IDE is where you author and review code in depth. A mobile client for managing agents is different — it is more like a job queue dashboard than a full coding environment. How Cursor develops the iOS app as the beta progresses — whether it stays lean as a task dashboard, or tries to expand it toward a fuller mobile coding experience — will tell us a lot about how this class of tool is meant to work.