Apple’s Siri roadmap moves from app controls to agentic goals

By Tom Whitfield

Apple expects two significant Siri improvements in the coming months: fine-grained control of first-party apps in iOS 27.3, followed by third-party actions likely to arrive in spring.

The sequence divides Siri’s development into two measurable interfaces: commands that alter content inside Apple apps and tasks that cross into external services. It also prepares a third stage, in which Siri pursues a goal across multiple steps rather than completing each request separately.

Bloomberg’s Mark Gurman reports CEO John Ternus is pressuring AI czar Craig Federighi to accelerate the roadmap and deliver an agent competitive with OpenAI’s Dots or Meta’s Muse. Federighi has already delivered a functional version of Siri AI on what Gurman calls an “incredibly short timeline,” although Apple does not consider itself to be keeping pace with rivals.

What changes for Siri users?

The iOS 27.3 release targets Apple’s own applications. Its two reported examples specify actions within separate interface contexts:

• Photos: Siri is expected to rotate and crop images.
• Safari: Siri is expected to organize tabs.

Fine-grained control here means Siri performs named operations, not merely offers guidance. For this stage, the measurable test is whether those operations execute successfully in the corresponding app.

The spring stage expands the task boundary from editing and organizing inside Apple apps to completing actions through outside services. The reported examples include ordering through Grubhub and booking a dinner reservation through OpenTable.

A spring arrival remains a forecast rather than a confirmed schedule. Gurman does not provide an exact date, a complete service list, or details about confirmation and error handling.

Mark Gurman said:

Dots, Muse and Grok can keep track of goals, determine what needs to happen next and continue pursuing objectives without the user initiating every individual step.

Apple ultimately wants to go in this direction, I’m told, and I believe the company could make Siri more like an agent as early as next year — not that it has much of a choice, given current consumer and market demands.

The quotation sets three conditions for the agentic stage. Each condition extends beyond a single command:

• It keeps track of a goal across actions.
• It determines what needs to happen next.
• It continues pursuing the objective without user initiation for every step.

Applied to the reported roadmap, the three stages have distinct success measures. iOS 27.3 is about completing named operations inside Photos and Safari; spring is about carrying an instruction into services such as Grubhub and OpenTable; the agent stage is about sustaining an objective and selecting the next action without repeated prompts.

For users, the practical change is fewer explicit instructions per completed objective. For Apple, the harder test is not whether Siri can execute one request, but whether it can decide and continue when an objective requires more than one action.

Gurman’s account leaves those harder mechanics unspecified. It does not state how Apple will define a goal, when Siri may act without another prompt, or how third-party integrations will be sequenced.

If Apple follows the reported sequence, iOS 27.3 will test app-level control and spring will test third-party execution. The decisive test by next year is whether Siri can carry a goal through those actions and continue pursuing it without a prompt for every step.