The Death of Icons: Why 80% of Mobile Apps Could Soon Become Unnecessary

# The Death of Icons: Why 80% of Mobile Apps Could Soon Become Unnecessary

The grid of apps on your smartphone screen won't last long. That is the suggestion from OpenClaw founder Peter Steinberger, who recently addressed Y Combinator investors with a provocative thesis: the operating system of the future doesn't need icons, it needs intent. And he is not exaggerating. By his estimate, eighty percent of existing apps will simply stop being opened by hand, because AI agents will open them for us, understanding what we need before we realize it ourselves.

It sounds like science fiction, but the mechanisms of this shift are already running. The industry is moving in three directions at once, and each of them is remaking what we have been calling the mobile interface. The first is a rebuild of the architecture at the logic level.

OpenClaw does not rely on a single neural network that answers questions. Instead, the project built a system that coordinates many specialized agents working in the background. Think of it as a swarm of bees, where each one handles its own task: one agent checks email, another books tickets, a third transfers money.

This distributed control architecture sits above individual apps: it doesn't call them, it orchestrates them. True, the success rate of multi-step operations still drops in more complex scenarios, but the principle is proven: a fundamentally new execution layer can be built on top of apps.

The second path is cruder but effective. ByteDance, the Chinese owner of TikTok, chose a strategy of visual capture. Rather than wait for developers to open up their apps' APIs, the company simply taught AI to look at the screen and "use" apps on its own, the way a person does.

It taps buttons, scrolls and types text, all through recognition of the visual interface. ByteDance has even launched its own smartphone with system-level AI for this purpose. The idea has shown its potential: the line between human control and automation is blurred.

But reality is harsher than the dream. With current mobile chipset performance (about 30 trillion operations per second), delays when switching between apps reach three seconds, and reliability barely clears the fifty percent mark. Still, the direction is set: hardware limits are temporary, architectural decisions are lasting.

The third and deepest shift is happening at the operating system level. Here the basic economic relations between the tech giants are being transformed. For ten years, Google paid Apple around two hundred billion dollars a year to make its results the default in the search bar. Search controlled the entry point to the digital environment. Now, as Bloomberg has reported, the situation is reversing: Apple may start paying Google for integrating Gemini, a neural network for system-level smartphone control. This is not just a change of partner; it redefines what an "entry point" is. If the entry point used to be the search bar, tomorrow it is a conversation with an intelligence built into the operating system itself.

For now, apps are not going anywhere. But their role is changing drastically. From points of interaction that the user deliberately opens, they become "digital pipelines" through which information flows at an AI's request. You won't open navigation, because the agent will already have called up the right coordinates; you won't search for flights, because the system already knows you need to be at the airport tomorrow at eight in the morning. The icon on the screen is an artifact of the past. The future doesn't need icons, because it speaks.