Report: Apple Inc. (AAPL.US) is developing an autonomous AI agent that can write code, browse the web, and control apps, competing head-on with OpenAI, Meta, and others.

date
06:58 20/11/2026
avatar
GMT Eight
Apple is developing AI agents for device control, apps, browsing, and coding.
Apple Inc. (AAPL.US) is developing its own autonomous AI agent that can get things done on users' behalftargeting a fast-growing market and competing with OpenAI, Anthropic, Meta Platforms and others. Apple Inc. outlined the plan in new job postings added to its website this week. In one of the postings, Apple Inc. said it is hiring a senior engineering manager for a "proactive, agentic AI assistant." According to the posting, the system will write code, browse the web, control apps and perform actions on devices on behalf of users. The move comes as Meta's Muse has become a hitin recent months, the product has captured consumer attention with its ability to book restaurants, order trips and prepare news briefs. Tools that let AI control Mac computers and write software are also proliferating. Anthropic recently expanded user testing of its own agent, Duet, while OpenAI expanded the capabilities of Dots. Alphabet's Alphabet Inc. Class C renamed and upgraded its agent to Tasks earlier this week. Like Elon Musk's Grok Bot, Tasks can place phone calls on users' behalf in addition to standard AI agent features. Apple Inc.'s App Store is already dominated by AI agents: as of Thursday, Muse had long held the No. 2 spot on the free app chart, behind only TikTok; OpenAI's ChatGPT and Anthropic's Claude also both ranked in the top five. OpenAI, Anthropic and Musk's SpaceXSI also rank in developer coding tools Codex, Claude Code and Cursor. Media reported earlier this month that Apple Inc. was developing its own AI agent service, and the new job postings provide fresh evidence of the company's plans. Apple Inc. has considered launching the service as early as the end of next year, but development remains at an early stage. An Apple Inc. spokesperson did not respond to a request for comment. "Imagine an AI agent that can use a computer the way a capable person does," the job posting reads. "It browses the web, works across apps, runs commands, writes and debugs code, calls services, and chooses the right tool at every step. It does all of this privately, with the kind of polish people expect from Apple Inc." The posting says the company already has a team building "the foundation for that future," and Apple Inc. wants the hired manager to "make it a reality." The tech giant's new CEO, John Ternus, has been urging software engineering chief Craig Federighi to pick up the pace on AI and take a bolder paththe agent is key to that effort. "The most valuable work people do today happens by moving fluidly between tools: the browser, a document, the terminal, an editor, a service. We are building an AI agent that can do the same," the posting says. It adds that Apple Inc. is developing agents that can "plan and execute multi-step tasks" across apps, browsers, command lines, code and application programming interfaces (APIs), and that they can also "coordinate specialized sub-agents along the way." The company strongly hints that its platform will be able to navigate autonomously across the interface of its own operating system. The posting explicitly mentions developing "general-purpose computer-use agents"the very technology that enables ChatGPT and Muse to control Macs. Alphabet Inc. Class C's Gemini also added such capabilities earlier on Thursday. Another core component is privacy. Apple Inc. says security needs to be built into the system design. "An agent that can run code, use credentials and act on a user's behalf needs strong permission models, sandboxing and transparency." Sandboxing refers to mechanisms that let engineers test software without affecting the external network, the rest of the operating system or the broader internet. While developing its own agent, Apple Inc. is also making it harder for rivals' Siasun Robot&Automation to gain broad control of its computers. In the upcoming macOS 27.3 beta, the company added a warning screen urging users to think twice before authorizing third-party agents to control their Macs. With the 27.3 changes, ChatGPT users now need to go through 14 different steps to enable support for third-party computer-use tools, up from just 4 previously. Apple Inc. frames the adjustment as a move to protect customers, but some AI companies and users see it as an attempt to slow competition. Apple Inc. has previously been sued by the U.S. Department of Justice over conduct including blocking third-party access to certain features and hardware. Apple Inc. is still far from launching a fully featured AI agent of its own, but it is taking steps to integrate the Siri digital assistant more deeply into third-party apps. The upcoming iOS 27.3 lays the groundwork for deeper collaboration between Siri AI and external appsfor example, letting the assistant book a car or a restaurant. Still, the upgraded system falls short of a fully featured agentone that can autonomously execute complex tasks across multiple apps and services in the background with little additional instruction from the user.