The consensus among analysts seems to be that Apple’s AI features announced at WWDC yesterday are largely catch-up with third party services. Yet Apple’s approach to AI is different. Instead of merely providing a chat bot for you to play with in the cloud, they are using the data on your device to personalize your AI. Instead of being limited to web results and its own corpus, Apple Intelligence will be able to return and interact with your data.

Why It Matters

Existing AI tools are largely deployed as fancy search engines. While useful, it is a large constraint. I have been impressed with the ability of the GPT-4o model to be hydrated with a large context window, specifically my SQL database scheme using TablePlus. Apple’s next generation AI tools will create new incentives to stay within the Apple ecosystem for its customers.

AI Forces a Cloud Retreat

Of note is the fact that supporting all its new AI features required Apple to build and deploy massive computing resources in the cloud. This likely explains why they are limiting the Apple Intelligence features to iPhone 15 or newer and MacOS users. The question is whether this is a permanent shift in how Apple provides services, or if improvements in software and hardware will allow it to provide these features entirely on device.

Zoom Out

With the unveiling of math notes and grammar checking Apple has mercilessly Sherlocked the Soulver and Grammarly apps. They never provide a reason why, but it seems like they think they can either do things better or like the feature but are un-impressed with the product direction and development.

One More Thing

With the arrival of Code Completion and Swift Assist, building apps in the Swift programming language may become easier than ever. This makes me wonder whether this will change the value proposition of using cross platform frameworks like Progressive Web Applications, React Native, and Electron versus maintaining a native Swift app for Apple users.