It is reported that Apple plans to upgrade Siri into an AI agent that can autonomously carry out tasks
According to Bloomberg reporter Mark Gurman in his Power On newsletter, Apple CEO John Ternus is pressuring software engineering chief Craig Federighi to accelerate Siri’s development roadmap, bringing Siri closer to an AI agent that can autonomously track goals, determine the next step, and keep pushing tasks forward, in order to compete with products such as OpenAI’s Dots and Meta’s Muse.
Gurman said he was told that Apple ultimately wants to move in this direction, and he believes the company could make Siri “more like an agent” as early as next year. At present, Siri still mainly relies on a request-response model; by contrast, systems such as Dots, Muse, and Grok can continuously track goals and keep work moving without users having to issue instructions step by step.
The report also said that Apple will first advance two related capabilities: one is more granular control over Apple’s own apps, expected to launch with iOS 27.3, such as rotating or cropping images in the Photos app and organizing Safari tabs; the other is integration with third-party apps, expected to arrive next spring, such as placing orders through Grubhub or booking restaurants through OpenTable. These are still on-demand triggered functions, while the agent capability that is closer to continuous autonomous operation remains a later goal.
Apple has not made any formal announcement regarding the related agent features, and the above timeline is Gurman’s judgment based on sources.
Single English source:
The Verge (November 8, 2026, citing Bloomberg’s Mark Gurman): Apple is racing to make Siri more agentic
ABAB AI Insight
As a matter of fact, this information comes from an unverified report by a Bloomberg reporter, and Apple has not publicly confirmed the specific timeline or scope of features; what has already launched or is about to launch is mainly in-app actions and third-party integration, rather than a fully autonomous continuous agent. Interpreted broadly, this reflects Apple’s internal intention, under pressure from its new CEO, to evolve from “passive response” toward “goal-oriented, multi-step autonomous execution” in order to close the gap with competitors in agent capabilities. The specific form of implementation, permission-control approach, and actual release timing remain uncertain.
Cognitive law:
When a voice assistant shifts from “single-command execution” to “continuously tracking goals and autonomously carrying out multi-step tasks,” its capability boundary expands significantly, but the requirements for user authorization transparency, privacy control, and interruptibility rise in tandem — the more a technology can do, the more clearly users must be able to know and manage what it is allowed to do.