Apple appears to be joining the computer-use agent race on its own terms. A job listing posted by the iPhone maker seeks a senior engineering manager to oversee the creation of an "AI agent that can use a computer" — evidence, first reported by Bloomberg, that Apple is developing an in-house autonomous agent capable of operating apps, browsing the web, writing code and calling services on a user's behalf.
The listing describes the ambition in unusually vivid terms for an Apple posting. "Imagine an AI agent that can use a computer the way a capable person does," the posting reads. "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."
According to the posting, Apple expects its agents to move "fluidly between tools" and to "carry out multi-step tasks across apps, browsers, command lines, code, and APIs." TechCrunch reported the listing on November 19 and said Apple did not respond to a request for comment.
Apple Enters a Crowded Agent Field
Computer-use agents — models that operate a desktop the way a person would, clicking through apps and typing into terminals — have become one of the most contested product categories in the industry. Anthropic and OpenAI have already launched such capabilities, and Meta recently entered the space with its Muse agent line. Apple, which has been widely criticized for moving slowly on generative AI, now appears ready to compete directly.
The listing lands in what TechCrunch calls the "new John Ternus era" at Apple, as the company works to expand its AI offerings. Bloomberg notes that Apple "has eyed introducing the service as soon as the end of next year," though progress on the in-house agent "remains in the early stages."
Tightening the Doors Behind It
The more provocative part of the report is what Apple plans for everyone else's agents. According to Bloomberg, the upcoming macOS 27.3 release will deploy warnings that discourage users from granting full system access to third-party AI agents. The move follows an October notification to developers, in which Apple said 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 dual-track approach — build your own agent while making rival agents harder to deploy — has drawn comparisons to Apple's historical playbook with the App Store. Critics will note that an agent with deep system access is both a competitive threat and a genuine security risk, which gives Apple a defensible rationale for the new warnings. Developers of third-party agents, meanwhile, may see the explicit-consent requirements as a barrier dressed up as a safeguard.
Why Computer Use Matters
The stakes are considerable. Agents that can reliably operate a computer would transform how people interact with software: instead of navigating menus, users could state an outcome and let the agent handle the clicks, keystrokes and API calls. That is why Anthropic, OpenAI, Meta and now reportedly Apple are all converging on the same target, and why "agentic" has become the industry's defining buzzword of the past year.
For Apple, the differentiation pitch is privacy. The job listing's emphasis on doing "all this privately" is a pointed contrast with cloud-first rivals, and consistent with Apple's on-device processing strategy. Whether an agent sophisticated enough to run a Mac can truly keep sensitive operations local remains an open question — long-horizon computer tasks have so far demanded frontier-scale models running in data centers.
There is also a timing question. Bloomberg's reporting suggests an end-of-2027 target at the earliest, which would put Apple at least two years behind Anthropic and OpenAI in the computer-use category, in a field where capabilities are compounding quarterly.
What to Watch
For now, everything about the project comes from a single job listing and unnamed sources, so details should be treated as provisional. Apple has not confirmed the agent, and the company routinely files for technologies that never ship. But the simultaneous move on macOS 27.3 — adding warnings exactly when third-party agents are proliferating — signals that Apple sees agent access to the Mac as a strategic chokepoint it intends to control.
If the listing's language is any guide, Apple's pitch will be an agent that is capable, private and polished. The harder question, which the company has not yet answered, is whether anyone else's agent will be allowed to be any of those things on a Mac.
Stay Ahead of AI
The AI agent race is accelerating across every major platform. For breaking AI news, in-depth analysis and coverage of every major model release, visit AI Buzz Wire — your front row seat to the intelligence revolution.
Read more AI news →