On 6 October Cursor added Remote Control to its iOS app. Developers can now view, and reply to, AI coding agents running on their own computers from a phone. The machine has to stay on and online, and Cursor’s release notes say the feature is on by default for everyone except Enterprise organisations.
It lands in a month of steady change at Cursor. Its Projects beta (10 September) introduced a coordinator agent that hands work to sub-agents and can run on schedules or react to Slack and pull request activity. A 23 September update cut token costs by 7% and added a Security Reviewer for pull requests on Teams and Enterprise plans.
What the trend actually shows
The unit of work is changing. A year ago a developer typed prompts and watched the result. Now an agent runs for hours, and the human’s job is to answer its questions, approve its direction and check what it produced. Remote Control just makes that supervision possible from anywhere.
What it means if you commission software
- Work can progress while your team is away from their desks. That is a speed gain, but it means review quality matters more, not less.
- “On by default” is worth a question. If your code or customer data sits on a developer’s machine, ask how agent access is governed and who can see it.
- Agents asking for human input is a healthy sign. Ask any supplier how often their engineers intervene, and what gets reviewed before it ships.
So what
The tools are moving fast enough that the useful question is no longer “do you use AI?” but “who is accountable for what the agents produce?” Good AI-assisted development keeps experienced engineers in charge of design, review and testing, with agents doing the heavy lifting. See how we work on our AI-assisted development page, or start a project and we will walk you through it.