Apple is reportedly building an AI assistant that would do more than explain which buttons to press. It would press them. Bloomberg’s Mark Gurman reports that new Apple job listings describe an autonomous agent intended to browse the web, work across apps, run commands, write and debug code, and navigate devices on users’ behalf. Source
Don’t outsource your to-do list just yet. Bloomberg says development remains in its early stages, although Apple has considered introducing the service as soon as late 2027. An Apple spokesperson didn’t respond to Bloomberg’s request for comment. Source
The hiring pitch seeks a senior engineering manager and asks readers to imagine an agent “that can use a computer the way a capable person does,” according to the listing reproduced by 9to5Mac. The posting describes agents that choose tools for individual steps and coordinate other, specialized AI agents on multistep jobs. Even the proposed assistant gets to delegate. Source
The listing mentions “general-purpose computer-use agents.” Bloomberg interprets that language as pointing toward software capable of autonomously navigating operating-system interfaces, a broader ambition than simply triggering predefined app functions. Source
That scope also goes beyond the Siri improvements Bloomberg describes for the upcoming iOS 27.3 release, which lays groundwork for deeper integration with third-party apps, including ride bookings and restaurant reservations. MacRumors reports that the role would work across Siri, UI frameworks and Apple Foundation Models teams. The broader agent’s final packaging and branding remain unclear. Source Source
Apple’s listing also calls for safeguards around that access. “Agents that can run code, use credentials, and act on a user’s behalf need strong permission models, sandboxing and transparency,” it says, as quoted by Bloomberg. Sandboxing limits software’s access to the rest of the system. Giving software the controls makes the permission screen rather more than a formality.
Comments