Why Siri's Big Rebuild Took So Long, and Why That Might Have Been the Right Call

For two years, the most reliable Siri story was a story about waiting. Apple promised a more personal, more capable assistant, then quietly admitted the work was taking longer than expected, then watched a leadership change unfold in public.

Now that the new Siri AI has arrived in iOS 27, Apple has finally offered an explanation for the delay. It is more revealing than the usual corporate shrug, and it reframes the wait as a choice rather than a stumble.

What Apple said

The explanation came from Mike Rockwell, who took over Siri last year after Apple moved him from the Vision Pro team amid a faltering Apple Intelligence rollout.

Speaking at a press session following the WWDC keynote, alongside Craig Federighi and Amar Subramanya, Rockwell described two different paths Apple could have taken.

The first path already worked. Apple had built an incremental version of the assistant, layered on top of the original Siri, that added tool calling. It functioned. But the team did not feel it delivered the experience they were after.

The second path required far more extensive changes, and Apple chose it instead. In Rockwell's telling, the company "tore it to the ground, rebuilt it from the ground up," this time on top of Apple's newer models.

The result, he said, is an assistant with its own application, natively multimodal, built with privacy from the start, and consistent across iPhone, iPad, Mac, Apple Watch, Vision Pro, CarPlay, and AirPods. One Siri, the same everywhere.

The choice underneath the delay

It is worth saying plainly that this is a more honest account than Apple usually gives about a missed timeline. The company is essentially admitting it had a shippable product and shelved it.

That decision deserves more credit than the long wait it produced has earned. Apple could have shipped the incremental Siri, claimed an Apple Intelligence win, and bought itself a year of better headlines.

The features would have been real. They also would have been a dead end, an assistant whose architecture could not grow into what Apple actually wanted.

Choosing to tear down working software because it was the wrong foundation is the kind of discipline that is easy to praise in hindsight and very hard to exercise when the clock and the critics are both running.

The reality is more complicated than a simple redemption arc, though.

The delay was not free. Apple set expectations early, with marketing that implied a personalized Siri was nearly here, and then spent two years not delivering it while competitors shipped. The leadership reshuffle that brought Rockwell in was not the move of a company executing smoothly. So the rebuild can be both the right call and the cleanup of an earlier overpromise.

Both things are true.

This is not the first time

Apple has torn something down and started over before, and the comparison that comes to mind is Maps.

When Apple Maps launched in 2012, it was a public embarrassment, and the fix was not a patch but years of quiet, fundamental rebuilding until the product became genuinely good. The lesson Apple seems to have internalized is that a shaky foundation is more expensive than a delay.

There is a difference this time, and it cuts against Apple's usual instincts.

The new Siri leans on models developed in collaboration with Google, even as Apple is careful to say Siri is not simply Gemini wearing an Apple badge. For a company that has long preferred to own its core technologies outright, leaning on an outside partner for the brains of its flagship assistant is a notable concession. It suggests Apple decided that getting Siri right mattered more than getting Siri entirely in house, at least for now.

What it means

If the new Siri lives up to the rebuild, the two lost years will fade into the kind of footnote that only enthusiasts remember. If it does not, the decision to discard a working version will look very different, and the explanation Rockwell offered will read less like discipline and more like a costly bet.

My read is that the choice was defensible, even admirable, and also that Apple does not get to take full credit for cleaning up a mess of its own making.

The company waited too long to promise and then had to wait again to deliver. But faced with a choice between shipping the wrong Siri on time and the right Siri late, Apple picked late.

For a product meant to anchor the next decade of its platforms, that is probably the call it should have made the first time.