Apple’s best AI idea looks a lot like vibe coding
Original reporting by The Verge

Apple's recent AI announcements at WWDC largely mirrored the industry's existing offerings: chatbots, text summarization, and image generation. Many features felt like a hurried attempt to catch up, simply porting familiar AI capabilities to the iPhone. Yet, within the early developer beta of iPadOS 26, a distinctive — though currently imperfect — vision for Apple Intelligence began to emerge, hinting at where the company could genuinely innovate.
A unique approach The core idea centers on making devices more adaptable to individual users through natural language. Tools like Shortcuts, traditionally powerful but complex, are being reimagined with AI. Users can now simply describe a desired action — "Send a text to Anna with three kissy emojis" — and the system attempts to build the automation. This "vibe-coding" approach allows users to sculpt their device's behavior without intricate programming, whether automating cross-app workflows or even creating simple browser extensions.
However, this ambitious vision is currently far from polished. Initial tests reveal significant bugs, with many complex natural language requests failing or requiring developer support. The broader success of this "agentic AI" hinges on widespread third-party developer adoption, a considerable challenge given that seamless automation might reduce direct app engagement. Despite these hurdles, Apple's unparalleled access to its device ecosystem and user data positions it uniquely. If it can overcome the technical and collaborative obstacles, Apple Intelligence could transform the iPhone not by reinvention, but by simply making everyday interactions profoundly easier and more personalized.
Apple's initial AI rollout, while often appearing to play catch-up with industry trends, reveals a more profound strategy focused on leveraging natural language to fundamentally enhance the existing user experience. Rather than introducing a revolutionary new interface, Apple aims to make its devices profoundly more personal, turning complex, multi-step tasks within Shortcuts and browser extensions into simple spoken or typed commands. The journey from a promising beta to a fully functional system is undeniably long, fraught with technical challenges and the critical need for widespread developer adoption, but the underlying principle is clear: empower users to mold their devices to their unique habits, minimizing friction and maximizing efficiency.