Apple promised us a smarter, more powerful Siri at WWDC 2024. The pitch was compelling: a Siri that understands your personal context, crawls through your messages and emails, carries out actions in your apps, and becomes a true assistant.
Two years later, that dream still remains a dream. But here’s the thing that could change the course of Apple’s assistant. Siri is reportedly no longer tied to a single AI brain. Apple is building it to be flexible and able to route requests to the external model that best does the job.
That made me ask a question. If Siri can use any AI, which one should she use? Currently the default external model is ChatGPT. But I would argue that twins are the more logical choice, and here’s why.
Siri is a search engine
Think about how you actually use Siri on a daily basis. They ask about the weather of the day. They ask about the nearest restaurants in your area. You ask it to look things up on the Internet. A significant portion of Siri usage involves searches or search-like queries, and no company in the world performs searches better than Google.
Google has spent decades building the most powerful search engine, and that expertise is now being incorporated directly into Gemini. When you ask Gemini something, she doesn’t just follow a language model. It extracts data from Google’s real-time web index, Google Maps, Google Shopping, and more.
Using this to improve Siri’s search functionality will help you reach new heights that no other LLM provider can match.
Apple promised personal information, but Gemini delivers
One of the key talking points in Apple’s WWDC 2024 announcement was personal intelligence. Apple has shown how Siri will display contextual information from all your apps and answer questions like “When does my mom’s flight land?” answered. or “Show me photos of Stacy in her pink coat from New York.”
In demo form it was really impressive. However, when I ask it to show me a photo of me in a black t-shirt, it brings up random photos of people from the internet wearing black t-shirts. I’m not exaggerating when I say that Siri’s personal intelligence feature was a colossal failure.
Meanwhile, Gemini has quietly launched its own personal intelligence feature. It accesses your Gmail, Calendar, Google Photos, Drive, and more to analyze your personal data and answer complex, life-contextual questions. It’s not perfect, but at least it works.
This is almost literally what Apple demonstrated as a future Siri skill, except Gemini is doing it today. If Apple wants to speed up the delivery of these features to users, Gemini may be the shortcut they need.
Gemini is already doing what Siri promised
Apple Intelligence delivers a compact, powerful AI model for all system applications, combining on-device processing to protect privacy with cloud-based computing for more demanding tasks. The on-device processing and data protection aspects are what set Apple apart from the competition. But it’s not alone now.
Gemini Nano already does this on Pixel and Samsung Galaxy devices. It enables offline summaries, smart replies and contextual features without requiring an internet connection. On Pixel 9 and newer, Gemini Nano is multimodal and can process images, text and language directly on the device.
Apple is building on what Google has already delivered. Rather than reinventing the wheel, using Gemini’s existing Nano deployment as the basis for Siri functionality on the device would save Apple a lot of trouble and money.
Gemini’s creative toolbox is packed
This is where things get really exciting. Gemini is not just a text model. It has an entire creative ecosystem that Apple could leverage.
Veo allows video generation up to 1080p, with cinematic styles and clips longer than a minute. Google DeepMind’s Lyria takes care of music and audio generation. For images, Nano Banana (Google’s image generation service) recently received a major upgrade with improved text rendering, theme consistency, and support for any aspect ratio.
Apple recently launched its own Creator Studio, giving users access to creative tools for a fixed monthly subscription. If the company is serious about competing with the likes of Adobe, it needs to offer generative capabilities. Imagine Gemini already has all of these features and it would make perfect sense to integrate it into Apple’s creative suite.
The partnership already exists
This point is not sufficiently discussed. Google reportedly pays Apple around $20 billion every year to remain the default search engine in Safari. This is one of the most valuable distribution deals in the history of technology. The relationship between Apple and Google is deep, long-standing, and financially huge for both companies.
Expanding this relationship from “Google supports Safari Search” to “Gemini supports Siri’s AI capabilities” isn’t a dramatic leap. It’s a natural evolution of a partnership that covers half of what happens when you open a browser on your iPhone.
So which model would I stick with?
Claude is excellent for reading long contexts and for nuanced thinking. ChatGPT has a huge ecosystem and powerful coding and agent tools. Both work great as user-selected specialists. I personally use Claude on my computer.
But as the default engine under Siri’s hood? They are not the right choice. Gemini operates at the operating system level on mobile devices, understands search queries and personal context, exists in the nano form factor on the device, and is at the heart of Apple’s most important relationship with a technology company.
The parts are all there. This isn’t about whether twins could power a smarter Siri. The question is whether Google and Apple can negotiate a mutually beneficial deal. And if the rumors are to be believed, things could already be heading in that direction.
