Apple fixed Siri in iOS 27, but the AI world has already moved on

The new Siri AI in the iOS 27 beta does everything Apple said it would, and it’s genuinely impressive. It finally understands personal context, rifling through your photos, emails, and texts to pull up a receipt you can’t quite name or a driver’s license number from a saved picture. You can have natural, back-and-forth conversations, type to it, or point your camera at a restaurant bill to split it. It’s the assistant everyone wanted a decade ago, built on a foundation of Google’s Gemini models that Apple trained and refined into its own proprietary system for on-device and private cloud use.

But the launch lands with a thud. Being a great Q&A bot and app launcher isn’t a breakthrough anymore. While Apple spent years fixing a fundamentally broken product, competitors turned AI into agents that code software, complete multi-step tasks, and reason across your entire computer. Delivering a polished Siri now feels less like innovation and more like finally patching a decade-old bug. The bar for basic competence was so low that simply playing the correct song on request counts as a notable win.

Apple’s partnership with Google is the quiet headline here. This isn’t a white-label job; Apple used Gemini to train its own Apple Foundation Models, optimized for its silicon. That technical approach paid off in a deeply integrated, private assistant. You can ask Siri to brainstorm recipes from random fridge contents or find a screenshot of a QR code from weeks ago, and it just works without punting you to a web search.

The public gets its hands on this in September with the official iOS 27 release. The question isn’t whether Siri is good—it finally is. The question is whether “good enough” cuts it when the AI conversation has shifted from helpful assistants to autonomous agents. Apple closed a gap that shouldn’t have existed, only to find the finish line moved.