Apple finally fixed Siri. So why does it feel anticlimactic?
TechCrunch Sarah Perez ● Covered by 2 sources
Siri's new AI brain just landed in Apple's iOS 27 beta, and it actually works now. Trouble is, the rest of the AI world stopped waiting years ago.
Based on reporting by TechCrunch, Sarah Perez — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Apple's Siri overhaul showed up in the iOS 27 consumer beta this July, and after years of delays, it does what Apple originally promised. It understands personal context, pulls in broad world knowledge, and digs through whatever is sitting on your iPhone to answer a question. And yet the whole thing lands with a shrug instead of a bang, because the AI world Apple is stepping into now looks nothing like the one it fell behind in. Tools elsewhere are writing code, running multi-step tasks on their own, operating computers, remembering things, generating media. A chatbot that finally works right isn't the jaw-dropper it would have been a few years back.
That doesn't mean the new Siri is skippable. You can hold an actual back-and-forth conversation with it now, adjust how expressive or fast its voice sounds, type instead of talk, or open a dedicated app that remembers past chats. Ask it to find your last saved receipt without telling it where it's stored, and it finds it. It can read a driver's license number off a photo, or dig up a QR code you screenshotted weeks ago. Point the camera at something in the real world and ask about it, or use it to split a dinner bill with friends.
Siri now reliably plays the song, podcast, or audiobook you actually asked for on the app you actually meant — which sounds like a laughably low bar, and it is, but it's one Siri has tripped over for years. It can also act as a sounding board: refining a half-formed idea, explaining an unfamiliar topic, or suggesting a recipe built from whatever's rotting in your fridge. Ask when your favorite band is playing near you and it answers correctly, using your location to sort out the details.
None of this happened in a vacuum. Apple built the new Siri through a partnership with Google, using Gemini AI models to train and refine its own Apple Foundation Models, rather than just rebadging Google's technology. Those models run on Apple's own silicon and its Private Cloud Compute infrastructure, which is presumably why Apple felt comfortable finally shipping this.
Still, the overwhelming feeling is relief rather than excitement — like a stubborn bug finally got squashed rather than something new got invented. This is simply how Siri was always supposed to behave. The surprise isn't that it's impressive; it's that it took this long to just work. Everyone outside the beta program gets it when iOS 27 ships publicly, expected in September.
My take — AI-written commentary, not fact-checked reporting
Apple gets credit for actually shipping a Siri that does what it says, using Gemini under the hood to get there instead of stubbornly going it alone. But there's no prize for arriving on time to a race that ended while you were still tying your shoes — fixing a broken product isn't the same as leading a category, and Apple spent years letting rivals define what an AI assistant even is.
Read more about this at: TechCrunch