cursor for ios is here — cloud agents, lock screen progress, voice-to-agent

Cursor launched a native iOS app in public beta on June 29, 2026, available to all paid subscribers. the pitch is not that you will write production code on an iPhone. it is that the agentic coding loop — kick off a task, let an agent work for hours, review the result, merge — does not require you to be at a computer for most of it.

what the ios app actually does

you can launch always-on cloud agents directly from the app, the same way you would from the desktop: choose a repo, describe the task, start the agent. the agent runs in Cursor's cloud infrastructure, not on your phone. your phone is the control surface, not the compute.

from there you can monitor the agent's progress, redirect it if it goes off track, review its proposed changes in a diff view, and merge the PR when you are satisfied. you can also attach screenshots, review videos and logs, and talk to the agent by voice. model choice is the same as desktop: OpenAI, Anthropic, Google, and Cursor's own Composer model are all selectable.

if you have a desktop session already running, the iOS app can connect to it. you can hand off context between devices — start something at a desk, check in from a phone, redirect from an iPad.

ios live activities on the lock screen

the most iOS-native feature is Live Activities integration. when an agent is running, you get real-time status on your iPhone lock screen and Dynamic Island: file currently being edited, step in the task sequence, time elapsed. you do not need to open the app to know where the agent is.

that is more useful than it sounds. long-running agents — multi-file refactors, test generation sweeps, documentation passes — can take 15 to 90 minutes. Live Activities means you can go do something else and get interrupted only when the agent finishes or gets stuck, not by polling the app every few minutes.

the mobile-first AI coding thesis

most coding tools were designed for the assumption that coding happens at a desktop with a keyboard, a large screen, and sustained focus. Cursor's iOS app is a bet that the agentic shift breaks that assumption. if an AI agent handles the sustained focus part — writing, testing, iterating — then the developer's job becomes: describe, review, redirect, approve. that workflow fits a phone well.

the framing in Cursor's announcement is explicitly "build from anywhere." that is a different product philosophy than "Cursor on mobile." the iOS app is not a port of the desktop editor. it is a control plane for agents that run in Cursor's cloud.

honest read: the app is in public beta, and beta software has rough edges. the 75% off Composer 2.5 runs promotion that launched with the app ran through July 5 and is now expired. the app is available now to all paid plan subscribers (Pro at $20/month and Ultra at $200/month). Free plan users are not included in the current beta.