Apple might be a bit behind in the AI arms race, but that might be for the best. Today, at WWDC, the company announced it’s bringing Siri AI to its devices, and it looks to be one of the more refined, restrained implementations we’ve seen. Finally, there will be a dedicated Siri app with a conversational interface, on top of the usual ever-present Siri interface. And all of it is built on top of Google’s Gemini models.
The Siri/Gemini combo doesn’t come as a surprise, as the partnership was announced earlier this year. The implementation, however, is interesting. In the past, Siri has existed as an abstract layer you can invoke while using your device, but in iOS 27, you’ll also be able to open the dedicated Siri app to start a conversation, similar to how most other AI assistants work.
Since Siri AI is built on top of Gemini, a lot of the functionality will look familiar if you’ve been keeping up with what Gemini can do. You can use it to look up directions, find specific photos in your camera roll, summarize documents, or generate event plans—with, of course, the usual caveats about the errors that generative AI can introduce.
On Macs, the new Siri AI will be built directly into Spotlight, which Apple says will be able to differentiate between typical searches and an AI request. How well this works could be crucial, since trying to find a particular file by name should have quick results, but generative AI responses tend to take quite a bit longer.
Siri will also be able to use your phone’s camera to identify objects or look up information. For example, Apple demoed using Siri to find nutrition info by pointing the camera at food, or splitting a bill by looking at the receipt. Siri will also be able to help users write or proofread their own writing from anywhere you can enter text.
All this will roll out with iOS 27 when it launches later this year, though there are some pretty major exceptions. First, due to regulatory issues, Siri AI will not be available in the EU on iOS or iPad OS at launch. Similarly, regulatory hurdles prevent Siri AI from coming to any Apple devices in China at launch.
There’s also the problem of usage limits. Specifics are still scant, but some Siri AI features will use on-device models, while others will rely on server-side models utilizing Apple’s Private Cloud Compute. However, the latter can take a lot of computing power, so some features like image generation will be subject to daily usage limits.
Apple hasn’t offered specifics about those daily usage limits, or a comprehensive list of which features might contribute to them. The company has said that iCloud+ subscribers will have “increased access,” though. So, that’s encouraging.
What do you think so far?