Carl Pei, the co-founder and CEO of Nothing, believes the next big leap in smartphones won’t be a slightly better camera, a faster chip, or another redesign of the same familiar home screen. In his view, the real disruption is far more fundamental: a future where apps fade into the background and AI agents take over the work people currently do by tapping through endless menus.
Speaking during an interview at the SXSW conference in Austin, Pei laid out a bold prediction for the direction of consumer tech. He argues that today’s app-centered model is heading toward a major shake-up, especially for startups whose value depends entirely on getting users to download and open a standalone app.
According to Pei, people should prepare for a world where “apps are going to disappear.” His point isn’t that software stops existing, but that the way we interact with software is likely to change dramatically. Instead of humans doing the work of navigating from app to app, AI agents would handle tasks directly, using services and tools behind the scenes without needing users to think in terms of “open this app, then that app.”
Pei has been discussing an AI-first device concept for some time, and this vision also helped Nothing secure a $200 million Series C funding round last year. The pitch centers on building a new kind of smartphone experience powered by AI and personalization that’s reliable enough that people don’t feel compelled to double-check everything it produces. At SXSW, he expanded on what it will take to move from today’s early AI features to something that truly replaces the app-driven workflow.
He described an early step already being explored in the industry: AI that can complete single commands on a user’s behalf, such as booking a flight or reserving a hotel. But Pei dismissed that stage as “super boring,” suggesting it’s only a small part of where AI assistants and AI agents are ultimately headed.
Where things start to get compelling, he said, is when an AI system begins to learn a person’s long-term intentions. Instead of reacting only to direct instructions, the device would understand ongoing goals and provide help automatically. If someone wants to get healthier, for example, the phone wouldn’t just wait for requests. It could offer useful nudges, suggestions, and timely prompts that support the goal over time.
Pei believes the most powerful shift happens when AI becomes proactive in a meaningful way. As the system gets to know users deeply, it can surface ideas and recommendations people didn’t even think to ask for. He compared the concept to the idea of AI memory, where the assistant gradually becomes more personalized and context-aware.
His broader frustration is that smartphones, despite all the technological progress inside them, still operate with an interface model that has barely changed in decades. Lockscreens, home screens, app icons, full-screen apps, and app stores still define the core experience. Pei argues that this structure is essentially a continuation of an older era of digital assistants and handheld devices, just refined and polished over time rather than reinvented.
In his view, this is why even simple real-world plans can feel surprisingly complicated on a modern phone. He gave an example most people recognize instantly: meeting someone for coffee. The intention is simple, but executing it often requires juggling multiple apps and steps—messaging someone, checking a location, arranging transport, and setting a time on a calendar. What should be effortless becomes a mini project scattered across different services.
Pei’s solution is an operating system that understands intention and acts on it directly. Instead of forcing users to manage the process manually through multiple apps, the phone would recognize what the person is trying to do and complete the workflow automatically through AI. The emphasis shifts from navigating software to getting outcomes.
That vision also comes with a major design implication: the interface itself would need to evolve. Pei argues the future isn’t an AI agent trying to use the same human-oriented interface—tapping buttons, scrolling menus, and mimicking touch. He sees that approach as a dead end. To make AI agents truly effective, devices should offer an interface built for the agent to use, not one that forces the agent to behave like a human operating a touchscreen.
At the same time, Pei acknowledged that apps won’t vanish overnight. The current ecosystem is too entrenched, and people still rely on app-based experiences every day. Nothing’s own software even supports the idea of users creating lightweight mini apps. But long-term, Pei thinks the direction is clear: the AI should be able to operate services frictionlessly, without the awkwardness of pretending to be a person clicking through screens.
Pei’s message is a challenge to the current smartphone status quo and a warning to anyone building a business that depends on a traditional app-first future. If AI agents become the main way people get things done, the winners may not be the companies with the most downloaded apps, but the ones whose services can be seamlessly accessed and executed by intelligent systems in the background.
For consumers, it’s a vision of a phone that feels less like a collection of icons and more like a personal operating partner—one that understands how you live, anticipates what you need, and does the tedious work so you don’t have to.
