According to a Bloomberg report on Thursday, November 19, 2026, citing job listings on
According to job descriptions such as "Senior Engineering Manager, Proactive Intelligent AI Assistant," the agent goes far beyond the capabilities of the current Siri and Apple Intelligence Suite, positioning it as a general-purpose application agent capable of operating a computer like a skilled human. Its core technical plans include cross-app operations, web browsing, code writing and debugging, running commands and making API calls, and the ability to plan multi-step tasks and coordinate specialized sub-agents.

To ensure system security, Apple explicitly requires building a strong permissions model, sandboxing mechanisms, and high transparency, continuing its long-standing on-device privacy protection strategy. The effort is led by software engineering chief Craig Federighi.
Currently, competition in the AI agent field is white-hot. Meta has risen rapidly with its Muse assistant and has long held the No. 2 spot on the Apple App Store's free chart, placing it in the top five alongside ChatGPT and Claude; Google also released a
Against this backdrop, while accelerating its own development, Apple has tightened control over its ecosystem. For example, in the upcoming macOS 27.3 beta, it added permission warnings and made enabling ChatGPT's computer control tool more cumbersome, increasing the steps from 4 to 14.
Affected by the news, Meta and Anthropic shares fell about 2% and 3%, respectively, in late trading. The move marks the full-scale start of a battle for the system-level entry point, with Apple trying to reshape the on-device AI competitive landscape by combining ecosystem barriers and its own agent.