Apple's rumored on-device language model could make autocorrect and predictive text much smarter, but it also raises questions about privacy, performance, and user control. In this episode, Lucas and Luna explore what a local large language model in iOS might mean for everyday typing, how it could learn from your writing style without sending data to the cloud, and the trade-offs Apple faces in making Siri and the keyboard feel more human. They dig into the possible hardware requirements, the role of the Neural Engine, and how this fits into Apple's broader AI strategy unveiled at WWDC 2026. If you've ever wondered why your phone still can't predict that you meant 'definitely' not 'definately,' this episode explains how on-device AI might finally crack that nut. Plus, they consider the competitive landscape, from Google's cloud-based models to Samsung's on-device approach, and what Apple's privacy-first stance might cost in terms of raw intelligence. For anyone who types a lot on an iPhone, this is a look at a feature that could quietly become a daily essential.