- Apple is developing an autonomous AI agent that could perform multi-step tasks across browsers, apps, code and device interfaces.
- According to Bloomberg, Apple is considering launching the agent in late 2027, although development remains at an early stage.
- The company’s job listing emphasized privacy, permission controls, isolated execution environments and transparency.
Apple is developing an autonomous AI agent and is considering a late 2027 launch as it seeks to compete with OpenAI, Anthropic, Meta and other companies in the agentic AI market. Bloomberg reported the plans, citing a job listing on Apple’s website for a senior engineering manager to work on a proactive AI assistant.
According to the job description, the agent would browse web pages, interact with apps, execute commands, write and debug code, and independently select the tools required to complete tasks. It would also plan and execute multi-step tasks across programs and services and coordinate specialized sub-agents when needed.
The job-posting page cited by Bloomberg was unavailable at the time of writing and said the position no longer existed or could not be accessed.
Apple outlines its agent plans
“Imagine an AI agent that can use a computer the way a capable person does,” the job listing 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.”
Apple’s work comes as competitors develop similar products. Meta is promoting its Muse agent, which can book restaurants, plan trips and compile news briefings. Anthropic has expanded testing of its Duet agent, while OpenAI is developing the always-on Dots agent and the Codex programming tool. Google has updated its Tasks agent, and Gemini has gained the ability to control Mac devices on users’ behalf.
An X user who goes by the name jack said agentic AI development would require more computing capacity than Apple currently has at sufficient scale. In his view, building infrastructure to process requests would increase demand for data centers, power supplies and construction permits, potentially benefiting AI infrastructure companies including IREN and NUAI.
Privacy and device controls
Apple’s job listing said agents capable of executing code, using credentials and acting on users’ behalf require robust permission mechanisms, isolated execution environments and transparency.
The company is also tightening restrictions on third-party AI tools seeking to control Mac computers. The macOS 27.3 beta added warnings that appear before users grant those apps the necessary permissions. ChatGPT users must now complete 14 separate steps to enable computer-control features, up from four previously.
Apple says the changes are intended to protect users, although some AI companies and users view them as an effort to make device control more difficult for competitors.
According to Bloomberg, development of Apple’s agent remains at an early stage despite the company’s consideration of a late 2027 launch.
Apple is separately working on deeper Siri integration with third-party apps. A future iOS 27.3 update is expected to prepare the assistant to perform specific actions, including ordering a car or booking a restaurant. The planned integration is not yet a full agent capable of autonomously completing complex tasks across multiple services.
Apple unveiled Siri AI in June as part of WWDC 2026.
Source: Incrypted
