Apple wants an AI that uses your Mac like you do

The job listing asks for a senior engineering manager to oversee the creation of what Apple calls an "AI agent that can use a computer the way a capable person does."

It is the clearest signal yet that Apple intends to compete in the agentic AI space, where assistants no longer answer questions but perform tasks across applications without human supervision.

The listing makes no mention of Siri by name, but the agent would likely build on Apple's existing AI infrastructure. Apple has been investing heavily in its AI capabilities since the company's leadership transition to John Ternus.

The new agent represents a significant expansion beyond the screen-awareness features Apple introduced with its AI-upgraded Siri earlier this year.

Bloomberg notes that at the same time Apple is building its own assistant, the company is making it harder for third-party agents to gain total control over its computers — a competitive move that could spark regulatory scrutiny given Apple's history of antitrust battles over its platform policies.

The company declined to comment when contacted by TechCrunch.

Apple is hiring a senior engineering manager to build an "AI agent that can use a computer" — an assistant that would browse the web, work across apps, run commands, write and debug code, and call services on behalf of the user.

Bloomberg first spotted the job listing on Apple's careers page. The listing describes an agent that "chooses the right tool for each step" and does everything "privately and with the polish people expect from Apple."

The engineer would work closely with teams across Siri, UI Frameworks, and Apple Foundation Models divisions. Apple's move reflects a broader industry shift toward agents that act independently rather than waiting for step-by-step instructions.

The job posting lays out a sweeping vision

Apple wants the agent to move "fluidly between tools" and "carry out multi-step tasks across apps, browsers, command lines, code, and APIs." The posting adds that agents should be able to coordinate "specialized sub-agents" along the way.

"People already do their most valuable work by moving fluidly between tools: a browser, a document, a terminal, an editor, a service," the job listing says. "We're building AI agents that can do the same."

These are not small ambitions.

An agent capable of chaining a browser search, terminal command, code edit, and API call is effectively a general-purpose computer operator — the same category as Anthropic's computer use feature and OpenAI's Operator product, but integrated directly into Apple's ecosystem.

Apple enters a crowded field of computer agents

The company's advantage may also be its tight integration between hardware, operating system, and first-party services. An agent built into macOS would have access to system-level APIs that third-party agents cannot match, even before Apple restricts their access further.

Meta's Muse and OpenAI's Dots rely on macOS accessibility permissions, which Apple can modify or restrict in a future update.

This gives the company a structural advantage in the agent race, similar to how its bundled apps dominate on iOS despite competition from third-party alternatives.

The question is whether Apple can build an agent as capable as the ones its competitors have already shipped.

Meta, OpenAI, and Anthropic have been iterating on computer-use agents for months, while Apple's listing describes a project that Bloomberg says remains in early stages.

Meta launched its Muse agent line this year, giving users an AI that can take control of a Mac through a dedicated app. OpenAI's Dots runs background tasks across applications.

Anthropic's Duet and Google's Tasks offer similar functionality. Perplexity also lets its assistant drive the computer through its Mac app.

Apple arriving late means the company will need to differentiate on privacy and ecosystem integration — two areas where the company holds an advantage.

The "privately" mention in the job listing is significant: Apple has marketed privacy as a core feature since the Tim Cook era, and an agent that sees everything on your screen raises obvious trust questions.

Apple blocks third-party agents while building its own

While Apple works on its own agent, the company is making it harder for competitors' agents to run on its hardware. Bloomberg reports that macOS 27.3 will display warnings discouraging users from granting full accessibility access to third-party agents.

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."

This dual strategy — building a first-party agent while limiting third-party access — mirrors Apple's approach to app tracking, where the company imposed strict limits on competitors while its own services were exempt.

Developers have already raised concerns about the fairness of the policy.

The Siri evolution and a long road ahead

Apple began rolling out AI-upgraded Siri in September, adding the ability to answer questions about what's on a user's screen and perform actions across apps.

Bloomberg reports Siri could soon rotate and crop images in Photos and organize Safari tabs — simpler steps toward the full computer agent described in the job listing.

With the upcoming iOS 27.3 update, Apple is also giving developers the ability to integrate Siri AI into "all apps" through a Developer Preview. The full computer agent remains in early stages.

Bloomberg writes that Apple has eyed introducing the service as soon as the end of 2027.

"Progress on its in-house agent remains in the early stages," the outlet says, suggesting the job listing is a first public step toward what could become the default way users interact with their Macs.

If Apple follows through, the agent would fundamentally change how users interact with their Macs. Rather than opening apps and navigating menus, users could describe what they want and let the agent execute each step.

That kind of shift carries risks: an agent with system-level access could introduce new security vulnerabilities, especially if a prompt-engineering attack tricks it into performing unintended actions.

Apple's emphasis on privacy in the job listing suggests the company is aware of these concerns and intends to build safeguards into the system.

The agent's architecture — running on-device rather than in the cloud — would align with Apple's privacy-first positioning and avoid the data-sharing concerns that have dogged cloud-based agents from OpenAI and Google.

But running a capable agent locally on a Mac also requires significant compute resources, which may limit the agent's capabilities on older hardware.

Common questions

What is Apple's AI agent?

Apple is developing an AI agent that can use a computer like a capable person. It would browse the web, work across apps, run commands, write and debug code, and call services, moving fluidly between tools.

When will Apple launch its AI agent?

The project is in early stages. Bloomberg reports Apple has eyed introducing the service as soon as the end of 2027, but progress remains early.

How does Apple's agent compare to competitors?

It would compete with Meta Muse, OpenAI Dots, Anthropic Duet, and Google Tasks. Meta, OpenAI, and Perplexity already offer Mac apps that can control a computer.

Apple is also restricting third-party agent access in macOS 27.3 while building its own.

Sources & further reading

  1. Apple may be developing its own AI agent that runs your Mac — TechCrunch ↗
  2. Apple is hiring engineers to build its own AI agent — The Verge ↗
How this story was made

Written by AI Lookout with AI assistance and checked against the linked sources. Company performance claims are attributed to the company. Analysis reflects AiLookout’s interpretation; we have not independently tested the products discussed. Cover photography is illustrative and does not depict the specific announcement or product.

Our editorial standards
Back to all stories