Apple seeks engineering leader for AI agent that works across apps
The hiring language describes software that would choose tools and carry out multi-step tasks. Development remains early, with no confirmed release date.
Loading page…
The hiring language describes software that would choose tools and carry out multi-step tasks. Development remains early, with no confirmed release date.
Listen to this story
Apple is staffing an early-stage effort to build an agent that can plan and carry out multi-step work across apps, browsers, command lines, code and APIs, including coordinating specialized sub-agents. Bloomberg reported that Apple has considered launching it as soon as the end of 2027, but that timing is tentative, not a product commitment. The immediate development is a search for an engineering leader; the listings describe intended capabilities and privacy goals, not a released system or verified safeguards.
Apple’s listing says the agent would choose which tool to use at each step, including websites, apps, commands, code and outside services.
The system is also envisioned as coordinating specialized sub-agents to handle parts of a task.
The listing promises private operation but provides no technical explanation of how user information would be protected.
Apple is hiring for an AI assistant designed to do computer work on a user’s behalf, not just answer questions. Bloomberg identified job listings describing software that would browse websites, control apps and write code. The November 19 disclosure offers a detailed view of the company’s ambitions, but not a finished product.
One opening seeks a senior engineering manager to oversee an agent that can use a computer, according to TechCrunch’s account of Bloomberg’s reporting. The role’s title, “Senior Engineering Manager, Agentic AI – Proactive,” places the work around an assistant that takes actions. Apple’s description says the team is building the foundation for that system.
The proposed assistant would move among several ways of getting work done. Alongside websites and apps, the listing names commands, software code and outside services. It also gives the agent responsibility for choosing which tool to use at each step, rather than describing a single-purpose tool for browsing or programming.
A passage reproduced by 9to5Mac sets out the intended experience in unusually direct language:
Imagine an AI agent that can use a computer the way a capable person does. 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.
Apple job listing, quoted by 9to5Mac
The scope extends beyond taking one action at a time. Apple describes agents that would plan and execute multi-step tasks across apps, browsers, command lines, code and APIs—the interfaces software uses to call other software. Those agents could also coordinate specialized sub-agents along the way, giving parts of the work to other agents.
Privacy is part of the stated goal, too. The listing says the agent would do its work “privately” and with the polish users expect from Apple. That wording describes an intended quality of the system, not a demonstrated privacy safeguard or a technical explanation of how the proposed assistant would handle information.
Bloomberg reported that Apple has considered introducing the service as soon as the end of 2027, according to TechCrunch. But the same account describes progress as remaining in the early stages. That makes the date a reported possibility, not an announced delivery commitment.
Apple did not respond to TechCrunch’s request for more information. For now, the concrete step is hiring an engineering leader to help turn the described system into a product—not making those capabilities available to users.
Loading discussion...
Join the conversation
Explain what would make that promise credible to you.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.