Why crafted apps will survive AI
There's a narrative gaining momentum in Silicon Valley — and it has spread across X like wildfire. It goes something like this: AI will replace apps. It will replace operating systems. There will be no more user interfaces. You'll just talk to an AI, and it will do everything for you. Elon Musk floated this vision on Joe Rogan's podcast. Countless threads on X echo it daily.
And then there are the startups building on this premise. "Create any app you can imagine." "Describe what you need, and AI builds it for you." The pitch sounds compelling. Infinite flexibility. No more settling for apps that don't quite fit your workflow.
I've been building apps for over a decade. And I think this vision fundamentally misunderstands how normal people interact with technology.
The imagination problem
"Create any app you can imagine" — that's exactly the issue. It assumes users want to imagine. That they walk around with a clear picture of the perfect tool in their head, frustrated only by their inability to code it.
They don't.
Most people don't think in terms of apps and features. They think in terms of problems. My baby won't sleep. I need a passport photo that meets official requirements. I can't hear conversations clearly anymore. They don't want to architect a solution. They want someone who already did that thinking for them.
When a sleep-deprived parent opens an app at 2 AM, they don't want a prompt. They want a button that plays the right sound, dims the right light, and lets them go back to bed. That's not a limitation — that's good design.
Shifting the burden
What the "AI replaces everything" crowd is really proposing is a massive transfer of cognitive labor — from designers and developers to end users. Today, when you open a well-crafted app, hundreds of decisions have already been made for you. Someone researched the problem. Someone tested different approaches. Someone decided what information to show first and what to hide behind a tap.
"Just describe what you need" reverses all of that. Now the user has to:
- Recognize they have a problem that software could solve
- Imagine what a solution might look like
- Articulate that solution clearly enough for an AI to build it
- Evaluate whether the generated result actually solves the problem
- Iterate until it works
That's not empowerment. That's a job. We used to call it product management.
People don't want to be creative about their tools
Here's what I've observed after shipping apps to hundreds of thousands of users: people are remarkably uncurious about their tools. And I mean that as a compliment. They have lives to live. A parent getting their baby to sleep, a tourist preparing a visa application, an elderly person trying to hear a conversation — these people are not looking for a creative exercise. They want the shortest path from problem to solution.
The promise of "generate any app" assumes a user who is technical, imaginative, and willing to iterate. That user exists — they're called developers. Everyone else just wants something that works when they open it.
Where AI actually helps
None of this means AI is irrelevant to apps. Quite the opposite. AI is transforming what crafted apps can do. Better personalization. Smarter defaults. Features that were impossible two years ago — like real-time speech-to-text, intelligent document scanning, or automatic photo validation.
The key difference: in a crafted app, AI works behind the interface, making the experience better without requiring the user to direct it. The user doesn't need to know it's there. They just notice that the app feels smarter, faster, more helpful.
That's the opposite of "talk to an AI and describe what you want." That's AI in service of design, not AI as a replacement for it.
Crafted means someone already did the thinking
When I build an app, the most valuable thing I bring isn't code. It's the hundreds of hours spent understanding a problem space. Talking to users. Testing assumptions. Making the hard decisions about what to include and — more importantly — what to leave out.
A well-designed app is an opinion. It says: "We studied this problem, and here's the best way to solve it." That opinion has value precisely because the user doesn't have to form one themselves.
"Create any app you can imagine" offers no opinion. It offers a blank canvas. And for most people, a blank canvas isn't freedom — it's friction.
The future is both
Will AI change how we interact with software? Absolutely. Will some tasks move to conversational interfaces? Sure — especially tasks that are one-off, exploratory, or genuinely unique.
But the idea that crafted apps will disappear? That we'll all be prompting our way through life? I don't buy it.
The apps that survive won't be the ones that resist AI. They'll be the ones that absorb it — using AI to become more intuitive, more personalized, more capable — while still offering what a blank prompt never can: a clear path from problem to solution, designed by someone who already did the thinking.
That's not a limitation of the old world. That's the whole point.