The Current AI App Gold Rush
Right now, the App Store feels like a digital gold rush for artificial intelligence. Dozens of apps offer AI-powered writing, image generation, or chatbot capabilities. For many users, these apps are their
primary window into the world of generative AI. But there’s a catch: a huge number of them are little more than fancy wrappers. They take a powerful model from a company like OpenAI or Anthropic, put a slightly different user interface on it, and charge a subscription fee, giving Apple its standard 15-30% cut. It’s a lucrative business model, but also a fragile one. Startups are essentially renting a storefront in Apple’s mall to sell a product made by someone else. This arrangement works only as long as the landlord doesn’t decide to start selling the same product itself.
A History Lesson in 'Sherlocking'
To understand where this is going, we need to talk about “Sherlocking.” It’s a term as old as the Mac itself, named after a 90s search tool called Watson that was made obsolete overnight when Apple released a nearly identical, free feature called Sherlock. For decades, this has been Apple's playbook: watch for popular third-party app categories, then build a native, free, and deeply integrated version into the operating system. We’ve seen it with flashlight apps, screen-recording tools, password managers, and widget platforms. The third-party app isn't banned; it's simply made irrelevant for the vast majority of users who will always prefer the built-in option. For AI startups, whose main differentiator is often just a slick interface on someone else’s technology, the threat of being Sherlocked isn't just possible—it’s probable.
The WWDC 2026 Playbook: Beyond the App
Apple's initial AI strategy, dubbed “Apple Intelligence,” is cautious. It’s about sprinkling AI features throughout the existing iOS experience. But by 2026, expect the integration to become far more profound and disruptive. The real danger for AI startups isn't that Apple will build a better ChatGPT app. The danger is that Apple will make the very *idea* of a standalone AI app feel obsolete. Imagine an AI so deeply woven into the OS that it’s simply ambient. Siri finally becomes the all-knowing assistant we were promised. Spotlight search can answer complex questions without opening a browser. Your photos, messages, and files are all indexed and understood by an on-device intelligence that third-party apps can’t access in the same way. In this world, why would you open a separate app and pay a subscription to summarize a document when you can just ask your phone to do it natively? Apple’s advantage isn’t just making a competing product; it’s controlling the entire operating system and user experience.
The Startup's Dilemma: Adapt or Disappear
So, how can an AI startup survive? The answer is to stop building things that Apple can easily replicate. The coming shakeout will separate the thin wrappers from the truly innovative companies. Survival will depend on building a “moat”—a defensible advantage that Apple can't or won't copy. This could mean focusing on a highly specific enterprise niche, like legal AI or medical diagnostics, where deep domain expertise is required. It could mean developing a proprietary AI model that is genuinely better than the alternatives for a specific task. Or it could mean building a strong community or network effect around a product that an OS-level feature can't replace. The startups that are merely putting a pretty face on GPT-5 won’t make it. The ones that are solving a unique, difficult problem for a dedicated audience just might. The App Store will no longer be a place to find a better AI chatbot; it will be a place to find specialized AI tools that do things the core OS doesn’t.
