Apple has posted job listings seeking a senior engineering manager to lead the creation of an AI agent that can use a computer on a user's behalf, according to reporting by
The Verge (theverge.com), which spotted the listings alongside Bloomberg. The role covers an agent that browses the web, works across apps, runs commands, writes and debugs code, and calls services, choosing the right tool for each step.
TechCrunch (techcrunch.com) reported that the posting describes an "AI agent that can use a computer" and that Apple did not respond to a request for comment.
The listings, added to Apple's website this week, are the clearest public evidence yet of a first-party agent strategy.
The Straits Times (straitstimes.com), citing Bloomberg, reported that Apple is seeking a senior engineering manager for a "proactive, agentic AI assistant" that will write code, browse the web, control apps and navigate devices. The same report notes that Apple has eyed introducing the service as soon as the end of next year, but that development remains in the early stages.
What the Listings Actually Say
The job text is unusually explicit about scope. "Imagine an AI agent that can use a computer the way a capable person does," the listing reads, according to both TechCrunch and The Straits Times. "It browses the web, works across apps, runs commands, writes and debugs code, and calls services, and it chooses the right tool for each step. It does all this privately and with the polish people expect from Apple." A second passage, quoted by The Verge and The Straits Times, frames the design goal around how people already work: "People already do their most valuable work by moving fluidly between tools: a browser, a document, a terminal, an editor, a service. We're building AI agents that can do the same." The listing adds that these agents "plan and carry out multi-step tasks across apps, browsers, command lines, code, and APIs" and can "coordinate specialized sub-agents along the way."
The Straits Times reports that the listing specifically mentions "general-purpose computer-use agents" and that the hired manager will work with teams across Siri, UI Frameworks and Apple Foundation Models, per The Verge. Privacy is presented as a design constraint rather than an afterthought: "Agents that can run code, use credentials, and act on a user's behalf need strong permission models, sandboxing and transparency," the listing says, according to The Straits Times.
A Crowded Field and a Closing Door
The timing matters because computer-use agents have moved from demo to product. The Verge notes that Meta's Muse, OpenAI's ChatGPT and Perplexity can already take control of a computer through their Mac apps, and that Google has just started letting Gemini do the same. The Straits Times reports that Meta's Muse has drawn consumer attention for making restaurant reservations, booking travel and preparing news briefings, that Anthropic has widened testing of its Duet agent, that OpenAI has expanded Dots, and that Google overhauled its agent, now called Tasks, earlier this week. It also notes that Muse held the No. 2 spot among free apps on Apple's App Store on 19 November, behind TikTok, with ChatGPT and Claude both in the top five.
Apple's own Siri has been moving in this direction. The Verge reports that Apple began rolling out its AI-upgraded Siri in September, which can answer questions about what is on screen and perform actions across apps, and that a Bloomberg report suggests Apple could add the ability to rotate and crop images in Photos or organise Safari tabs. The Verge adds that the upcoming iOS 27.3 update will let developers integrate Siri AI into "all apps" through a Developer Preview.
At the same time, Apple is tightening the perimeter around rivals. TechCrunch, citing Bloomberg, reports that macOS 27.3 will deploy warnings discouraging users from granting full access to third-party agents, and that in October Apple told developers it planned to "introduce additional controls to ensure that users who genuinely wish to grant an app this extraordinary level of access can only do so with very explicit user action." The Straits Times puts a number on the friction: with the change in 27.3, ChatGPT users now need to engage in 14 distinct steps to enable support for third-party computer-use tools, up from four. It notes that while Apple has positioned the shift as customer protection, some AI companies and users see it as an attempt to slow the competition, and that Apple has already been sued by the US Justice Department over actions such as blocking third-party access to features and hardware.
Why It Matters
For anyone building agentic tools on Apple's platforms, the strategic question has shifted. An OS-level agent with privileged access to apps, credentials and system interfaces is a different kind of competitor from a third-party app asking for the same permissions. Apple's own framing, with its emphasis on permission models and sandboxing, suggests the company intends to make the native agent the trusted path and third-party access the exception requiring explicit consent. That is a platform decision as much as a product one.
The commercial read-across is already visible in markets. The Straits Times reports that shares of Meta fell nearly 2% and Anthropic dropped more than 3% in late trading after Bloomberg reported on Apple's efforts, a modest but telling signal that investors treat a first-party Apple agent as a competitive threat rather than a neutral addition.
What Is Still Unknown
Much remains unresolved. Apple has not confirmed the product, and an Apple spokesperson did not respond to a request for comment, according to The Straits Times. The end-of-next-year timing is described by Bloomberg, via TechCrunch, as an ambition rather than a schedule, with progress "in the early stages." The Straits Times notes that the updated Siri integration in iOS 27.3 does not go as far as a full-fledged agent capable of autonomously carrying out complex tasks across multiple apps in the background. How Apple prices or bundles the agent, whether it runs largely on-device, and how it treats third-party agents that complement rather than compete with it are all open questions the listings do not answer.
FY Outlook
The direction of travel is clear even if the timetable is not. Apple is assembling the teams, the model stack and the permission architecture for a native agent, while using OS-level warnings to raise the cost of granting equivalent access to outsiders. The next credible checkpoints are the iOS 27.3 and macOS 27.3 releases and any further hiring or organisational signals from the Siri and Foundation Models groups. Until Apple ships something, the listings are evidence of intent, not of capability. But intent, backed by control of the operating system, is the part competitors should watch most closely.
Sources and References