Siri is no longer a voice assistant. Apple's planned overhaul crosses a structural threshold: transforming the primary interface on 2 billion devices from command-execution to conversational reasoning. The move, powered by a custom Google Gemini model and launching in September 2026, signals that major OS makers are now embedding conversational AI natively rather than directing users to standalone applications. This isn't a feature refinement—it's an architectural rebuild that rewires how computing intent flows through consumer devices.

The moment is here, and it's architectural. Apple isn't adding an AI feature to Siri—it's replacing Siri entirely. According to Mark Gurman at Bloomberg, the company's plans for iOS 27, launching in September 2026, will scrap the existing voice-command interface and rebuild it as a reasoning-based conversational chatbot. Code-named Campos, this version will debut at WWDC in June, becoming what Apple's calling the "primary new addition" to iOS 27, iPadOS, and macOS 27. Everything else in the OS update is focused on stability.

That distinction matters. Not a feature addition. Not a layer on top. The primary product change for what will run on roughly 2 billion devices is a new interface paradigm.

The power move here is the technical choice: Apple is building this on a custom Google Gemini model, deepening the partnership announced earlier this year. The chatbot will operate in multi-modal mode—typing and voice—matching the interaction model users have gotten accustomed to through ChatGPT, Claude, and other standalone conversational tools. This version will "significantly surpass" the AI personalization features arriving in the near term, Gurman reports. Translation: Apple is committing to reasoning depth, not surface-level intent parsing.

Here's what makes this a true inflection point and not just competitive catching-up. The transition from Siri-as-command-interface to Siri-as-reasoning-engine represents a platform-layer architectural shift. For fifteen years, Siri has been structured around command completion—"Set a timer," "Call Mom," "What's the weather." The interface expects discrete tasks. Conversational AI flips that premise. Instead, the interface now expects open-ended dialogue where context builds across turns. The system needs to maintain state, reason about intent across multiple sentences, and handle ambiguity. That's a fundamentally different application layer.

More strategically, this signals where platform makers have decided the AI distribution battle will be won. Google has been embedding Gemini into Workspace for months. Amazon Health AI is native to the device. Now Apple, the creator of the "app economy" that made standalone distribution the path to reach users, is saying: conversational AI is too central to user experience to be relegated to downloaded applications. It's moving to device core.

For builders, that's the inflection. If conversational AI is shifting from application layer to OS layer, the competitive moat isn't in the chatbot anymore—it's in either the underlying model (Google Gemini, OpenAI's relationship with other platforms) or the data integration (Apple's device context, Microsoft's enterprise integration through Windows). Standalone chatbot companies are facing structural headwinds. The attention is migrating to the system layer.

The timing is deliberate. Apple has spent the last two years watching OpenAI grow from zero to consumer ubiquity. ChatGPT showed the interface model. But OpenAI doesn't control distribution—it's dependent on App Store placement, browser defaults, and user habit to pull people to standalone apps. Apple, with two billion devices and a refresh cycle that forces user attention twice a year, can flip that. Starting in September 2026, conversational AI isn't optional or supplementary. It's the primary interface evolution.

What Apple didn't do is build this on proprietary silicon. The Gemini partnership suggests Apple made a calculation: speed-to-execution and reasoning capability matter more than vertical integration here. That's Google's win in the partnership. That's also a signal to enterprises that any ecosystem relying on proprietary model lockdown faces a harder competitive fight.

The execution details matter enormously and remain vague. WWDC in June will need to answer: How much reasoning happens on-device versus cloud? How does Apple handle fallback when Gemini's API is unreachable? What's the privacy story when conversational reasoning requires more data context than Siri historically needed? The difference between a conversational OS interface that works reliably and one that stutters is measured in milliseconds and error handling. That's a 6-month sprint to solve.

For decision-makers in enterprise, the implications are immediate. This is the moment where device-layer AI becomes mandatory to plan for. Your employees are about to have a device-native interface that reasons conversationally. How they'll use it, what data it can access, what policies govern it—those are now 2026 IT planning conversations, not 2027 discussions.

This is the moment the conversational AI interface transitions from novelty to OS infrastructure. Apple's architectural rebuild signals that platform makers have decided the battleground is device core, not application stores. For builders, the shift means conversational AI must now be OS-native to matter competitively—standalone distribution is becoming a feature layer, not a primary channel. Investors should watch how model partnerships reshape—Apple-Google's Gemini integration suggests reasoning capability and speed-to-execution matter more than proprietary vertical control. Decision-makers need to treat device-layer conversational AI as a 2026 IT planning mandate, not 2027. The next threshold: execution details at WWDC in June, particularly on-device reasoning capabilities and privacy handling for conversational context.