Apple unveiled a long-awaited AI-powered overhaul of Siri at its WWDC keynote on Monday, marking the first major revamp of the voice assistant in roughly two years and coming amid a pending $250 million lawsuit over its delayed rollout. The new features, branded under "Apple Intelligence," will roll out across iPhones, MacBooks, and the Apple Vision Pro headset. In a live demo, Justin Titi, Apple's senior director of AI engineering, asked Siri to remind him of a dessert his daughter had recently mentioned. The assistant searched through a month of text messages, located a reference to coconut cookies, and surfaced it on command, a small but illustrative example of how Apple is pitching contextual, cross-app awareness as the centerpiece of the update. Still, the author of the piece isn't fully sold. She writes that she doesn't trust large language models to deliver consistently accurate information, finds it ethically uncomfortable to use AI to help her write, and has no interest in generating stylized images of herself, like the viral Studio Ghibli trend. What she does want is a "second brain" akin to Emily, the sharp assistant from "The Devil Wears Prada," something that anticipates her needs before she articulates them. That means Siri parsing a text thread to automatically create a calendar entry when she and a friend lock in Thursday dinner plans, pinging her when she walks past CVS with a prescription waiting, or nudging her about a work email she forgot to answer. The WWDC demo offered a glimpse of that direction but stopped short of the fully autonomous assistant the author envisions. She acknowledges Apple's approach is "moving in the right direction," particularly the ability to search across apps and timeframes without manual scrolling, but she frames the entire pitch with a tension: the convenience of an always-on assistant that knows everything about your digital life is exactly what makes it feel invasive. As she puts it, paraphrasing Katy Perry, the privacy trade-offs feel wrong, but the relief of offloading cognitive load from a phone overloaded with a dozen active conversations feels undeniably right.