Apple’s transition to the new Siri in iOS 27 was delayed because engineers opted to scrap an incremental update in favor of a complete architectural rebuild. According to Mike Rockwell, who leads the Siri division, the company discarded a functional prototype that merely added tool-calling capabilities to the legacy system. Apple determined that the original design failed to meet the vision for a natively multimodal, privacy-focused assistant, necessitating a ground-up reconstruction of the software.
Why Apple abandoned the incremental Siri update
Apple chose to prioritize long-term performance over a faster release schedule. Mike Rockwell, speaking during a post-WWDC technical session, confirmed that his team had an operational version of Siri ready last year that functioned as an extension of the existing infrastructure. However, leadership decided this approach lacked the depth required for the new standard of AI integration. By choosing to “tear it to the ground,” the company moved to integrate Siri directly with the new foundational models described by Apple executive Amar Subramanya.
The shift to a natively multimodal architecture
The primary advantage of the new architecture is native multimodality, which allows Siri to process different types of information simultaneously rather than relying on disparate, bolted-on systems. According to Rockwell, this shift allows the assistant to operate with a unified experience across the entire Apple ecosystem, including Vision Pro, CarPlay, and AirPods. This marks a significant departure from the legacy Siri, which often functioned as a collection of siloed commands rather than a cohesive, intelligent agent.
What does this mean for privacy and cross-platform consistency?
By rebuilding the assistant from the ground up, Apple claims it has embedded privacy protections into the core of the system rather than applying them as a secondary layer. Rockwell emphasized that this new version is “privacy from the ground up,” a design choice that would have been difficult to implement if the company had proceeded with the earlier, incremental version of the software. This common foundation ensures that user intent and data processing remain consistent whether a command is issued to a HomePod or a MacBook.
Frequently Asked Questions
Why did Siri take so long to update?
Apple decided to rebuild the software from the ground up rather than releasing an incremental update that only added basic tool-calling features, according to Mike Rockwell.
Is the new Siri different on Mac compared to iPhone?
No. Apple designed the new system to provide a common experience, meaning the underlying logic and capabilities are consistent across iPhone, iPad, Mac, Apple Watch, and Vision Pro.
What does “natively multimodal” mean for the user?
It means the assistant is designed to handle multiple types of inputs and tasks simultaneously within a single, cohesive application, rather than relying on the older, more limited command structures.
What is your experience with the new Siri interface so far? Does the improved cross-platform consistency change how you use your devices? Share your thoughts in the comments below.
