Key facts
- Bloomberg reported on November 19, 2026 that Apple is developing an autonomous artificial intelligence agent to enter a market pursued by OpenAI, Anthropic PBC, Meta Platforms Inc. and others.
- The position is titled Senior Engineering Manager, Agentic AI – Proactive, as shown in the November 17 posting.
- The posting says the manager will define the technical vision, roadmap, and quality standard for AI agents operating across apps, browsers, terminals, code, and programming interfaces, as stated on November 17.
- Safety, privacy, and security must be central to the design of the agentic AI, according to the November 17 posting.
On November 17, 2026, Apple advertised a managerial role at its Cupertino campus to oversee a proactive artificial intelligence assistant built to act on its own.\n\nThe posting describes an assistant that browses websites, moves between applications, executes commands, creates and fixes code, calls services and selects the proper tool for each step. The ad tells applicants to imagine an assistant that runs a computer the way a skilled person would. It can plan multi-step tasks across programs, web browsers, command lines, software and APIs, then pass pieces of work to smaller specialized assistants. The assistant can write software, visit websites, take control of programs and move through devices on a person’s behalf.\n\nThe job is titled Senior Engineering Manager, Agentic AI – Proactive.
The person in the role will oversee the agent platform and AI capabilities, manage other managers and engineers, and collaborate with teams working on Siri, user-interface frameworks and Apple Foundation Models. The manager will set the technical vision, roadmap and quality bar for agents that work across apps, browsers, terminals, code and programming interfaces.\n\nApple wants at least 10 years of software engineering, including five years leading engineering teams, and experience supervising other managers. It also favors candidates who know sandboxing and permission settings for software that runs code or acts for users, who can shrink models using quantization and distillation for limited hardware, and who have built automation and accessibility tools or created simulation environments and benchmarks to test agents on full tasks.\n\nThe company says the assistant will work in private, and the posting says safety, privacy and security should sit at the heart of the agentic AI’s design.\n\nThe company has not confirmed the project, and an Apple spokesperson did not answer a request for comment. Stories on November 20, 2026 said Apple seems to have taken down the listing, and its careers page now states the position is unavailable.
Before it came down, AI commentator Andrew Curran posted a screenshot on X that matched the job number, the November 17, 2026 posting date and the text. The company has not said when it will ship or which devices will get it.\n\nBloomberg said on November 19, 2026 that Apple is building an autonomous artificial intelligence agent to enter a market chased by OpenAI, Anthropic PBC, Meta Platforms Inc. and others. The company revealed the plans through job listings placed on its website this week. Bloomberg said earlier this month that Apple was developing its own AI agent product, and the listings provide new evidence of the company’s plans.\n\nMeta’s Muse holds the number two spot among free iPhone apps, trailing TikTok, and it passed 10 million downloads within 40 days, according to Sensor Tower.
ChatGPT holds the number three spot, Claude is number five and Gemini is number seven among apps with agentic abilities, while OpenAI’s ChatGPT and Anthropic’s Claude both appear in the top five. Macworld calls proactive AI agents that run nonstop in the background the next frontier for consumer AI and the natural successor to chatbots.\n\nMeta put out its Muse agent in September 2026. OpenAI introduced Dots, Anthropic put out Claude Duet, and Google turned Gemini Spark into Tasks with new features. Google’s Gemini app for the Mac got an update allowing it to run a Mac for the user.\n\nGoogle updated its agent, now called Tasks, earlier this week; like Elon Musk’s Grok Bot, Tasks lets the system place phone calls for a person.
OpenAI, Anthropic and Musk’s SpaceXSI stand behind the developer coding tools Codex, Claude Code and Cursor. Anthropic PBC has recently widened user testing of its agent Duet, and OpenAI has added capabilities to Dots.\n\nBloomberg said Apple hopes Siri works like an agent by 2027, and stories on November 20, 2026 said Apple wants Siri to act more like an autonomous assistant, near Meta’s Muse and ChatGPT’s Dots. The iOS 27.3 beta includes Siri AI actions spanning 15 Apple apps, and Apple plans to release the update in December. But Siri AI in iOS 27 still waits for each request, while competitors ChatGPT Dots and Meta Muse can keep chasing a goal across conversations.
ChatGPT and Claude both offer Computer Use, a tool that looks at a screen, presses keys and types text into apps; Apple has no equivalent for iPhone, iPad or Mac.\n\nApple put in place new limits on how outside AI agents can reach and control Macs. The company said last month it plans to tighten macOS privacy protections around Full Disk Access, and the macOS 27.3 beta lets users control more precisely which apps and data outside AI agents can reach.\n\nTurning on ChatGPT control of a Mac in the macOS 27.3 beta takes 14 separate steps, up from four before. Apple says security is the reason, but some AI companies and users see it as a move to slow competition. Agents on phones and computers usually work together, coordinating tasks across devices and software to reach increasingly complex goals.\n\nMeta shares fell nearly 2 percent and Anthropic fell more than 3 percent in late trading after Bloomberg reported on Apple’s work on November 19, 2026.
Stories on November 19, 2026 said Apple has thought about launching the service before the end of next year. Mark Gurman thinks Apple could bring out these capabilities by next year. The work is still in its early stages.

Join the discussion