Apple Siri AI overhaul

OpenAI's Dots and Meta's Muse already run in the background, executing multi-step tasks and monitoring digital workflows while the user looks away. The critical difference is that Apple controls the underlying operating system those third-party agents run on. Now, the hardware giant is racing to build that same autonomous capability directly into the iPhone.

Apple is planning a fundamental overhaul of its upcoming iOS 28 software to turn its virtual assistant, Siri, into an autonomous AI agent by 2027. The directive represents a massive technical pivot for a company that has historically favored tightly controlled, request-based interactions over open-ended automation.

The mandate comes straight from the top of the corporate hierarchy. Apple chief executive John Ternus is reportedly pressuring software chief Craig Federighi to accelerate the Siri roadmap and move beyond incremental fixes. The goal is to match the continuous, background-running capabilities of rival systems before consumers permanently adopt third-party alternatives.[2][3]

The Agentic Shift

To understand the scale of this overhaul, it is necessary to distinguish between an assistant and an agent. Today's Siri is strictly request-based: a user asks for a timer, dictates a text message, or requests a weather update, and the software executes that single command before returning to a dormant state.

An autonomous agent operates on an entirely different architecture. It holds a specific goal across multiple steps and extended periods of time. An agentic system can track the status of a refund, monitor project updates across different applications, or assemble a weekly summary without needing a fresh prompt for every individual action.

According to a November 8 report from The Verge, Apple wants Siri to behave more like a proactive agent, moving closer to the functionality of ChatGPT Dots than the request-and-reply system most people use today. This requires an orchestrator-driven architecture that maintains deep personal context.[1]

This represents a massive technical leap that Apple has not yet achieved. Making an assistant genuinely agentic requires giving it persistent access to system functions and third-party applications. The software must be able to read the screen, understand the relationships between different apps, and execute commands without breaking security protocols.

Apple's leadership recognizes that its current capabilities lag significantly behind dedicated competitors. "While Federighi has delivered a functional version of Siri AI on an incredibly short timeline, the company has no illusions that it's keeping pace with its rivals," The Verge reported, highlighting the internal acknowledgment of the widening capability gap.[1]

Restructuring The iOS Factory

To support this ambitious transition, Apple is fundamentally altering how it builds its mobile operating system. The development pipeline for the upcoming iOS 28 has been restructured to prioritize complex feature exploration over rigid release schedules, a rare move for the highly regimented engineering organization.

Previously, Apple software engineers worked on a strict cycle of four weeks for coding and four weeks for integration. That rhythm favored stability and predictable bug-fixing, but it left little room for experimenting with unpredictable artificial intelligence models that require extensive tuning.

Under the new pipeline, the first iOS 28 milestone gives software engineers six weeks for coding and feature development, followed by just two weeks for convergence. This extended runway allows teams to test agentic behaviors and build parallel prototypes before locking features into the main operating system branch.

This structural change in how iOS is built highlights the technical difficulty of creating an autonomous agent capable of executing complex tasks. By extending the development runway, Apple aims to give its engineering teams the necessary flexibility to refine these advanced capabilities prior to their public release.

Employees reportedly favor the pipeline adjustment. The additional two weeks of coding time provide engineers with the breathing room needed to explore dead ends and refine machine learning models without the immediate pressure of merging their code into the master branch for stability testing.

The Internal Push

The urgency behind this pipeline shift stems from a closed-door meeting between Ternus and Federighi. Ternus expressed frustration over what he perceives as a reluctance to embrace artificial intelligence with the necessary speed, pushing for a more aggressive posture against industry competitors.

"Apple CEO John Ternus is reportedly pushing software chief Craig Federighi to speed up the company's AI work and build a Siri that can act as an autonomous agent," MLQ.ai reported on November 8. The internal pressure underscores a shift in corporate culture following Ternus's recent elevation to the chief executive role.[2]

Federighi has long been regarded as somewhat of an AI skeptic within the industry. He has frequently raised concerns about the technology's implications for user privacy, system security, and broader societal impacts, favoring a cautious approach that keeps data processing strictly on the device.

Despite those reservations, Federighi's software engineering group managed to deliver a functional version of Siri AI on an incredibly short timeline this year. However, that release merely modernized the assistant's conversational abilities; it did not introduce the autonomous, background-running features that define the current frontier of generative AI.

Ternus is now pushing the software chief to pursue a more ambitious overall strategy. The internal pressure highlights the inherent tension between Apple's traditional privacy-first, on-device processing model and the massive computational requirements of autonomous agents that typically rely on cloud infrastructure.

The Stepping Stones

Before Siri can become a fully autonomous agent in 2027, Apple is laying the necessary groundwork with a series of intermediate software updates. Two significant improvements are expected to roll out to consumers in the coming months, establishing the technical foundation for agentic behavior.

The first milestone, expected in the iOS 27.3 release this December, will grant Siri fine-grained control over Apple's first-party applications. This update moves the assistant beyond simple dictation, providing it with direct programmatic handles to manipulate media and browser workspaces.[1]

This update will allow the assistant to execute precise operations entirely through voice commands. Users will be able to ask Siri to rotate and crop images inside the Photos app, or to restructure and organize tab groupings within the Safari web browser.[1]

Following that initial release, a broader programmatic shift is planned for the spring of 2027. Apple intends to open these automated execution pathways to third-party services through its App Intents framework, allowing external developers to plug their software directly into Siri's orchestration layer.

This spring update will enable Siri to execute multi-parameter transactions across external platforms. In practical terms, this means the assistant could book dining reservations on OpenTable or place food delivery orders through Grubhub without requiring the user to manually open either application.[1]

The Ecosystem Stakes

These intermediate steps are crucial for building a machine-readable action surface. An autonomous agent cannot function if it cannot reliably interact with the software installed on the device; it needs developers to explicitly declare what their applications can do.

AI

By asking developers to map their app's capabilities through App Intents, Apple is creating the execution environment that a future agentic Siri will rely upon. The companies that publish clean intents early will be the ones the agent can operate smoothly.

The stakes for Apple are incredibly high. The assistant layer of the operating system is widely viewed by industry analysts as the next major battleground for user retention and ecosystem lock-in. The company that controls the agent controls the primary interface through which users interact with the digital world.

If iPhone users begin defaulting to third-party interfaces like OpenAI's Dots or Meta's Muse to navigate their daily tasks, Apple risks losing its position as the primary orchestrator of the mobile experience. The underlying hardware would become a commoditized layer for someone else's software.

The iOS 28 overhaul serves as a stark acknowledgment that a mobile operating system without native, autonomous AI orchestration will soon be viewed as functionally obsolete. Apple is betting that a restructured development pipeline and a clear mandate from the top will be enough to close the gap.