Scroll through the feed these days and you'll see the same story. AI coding agents are getting better every week. Software is becoming commoditized. Writing code is increasingly banal. If you're thinking about starting a software business, you've already missed your chance.
The barriers have fallen. The code writes itself. The market is flooded.
But something's missing from this story.
The story assumes that software is code.
It assumes that the hard part of building a tool is writing the code to make it work. It assumes that once you can generate apps in minutes, what makes something worth building collapses.
I tried this. I sat down with a clear plan and a vision, and I used AI to generate what I asked for. Feature by feature, component by component. And the code worked. It was impressive.
But the tool didn't feel right. It felt hollow. Like a house with all the furniture in the right rooms but nobody lives there.
Because code is not software. Code is just the artifact.
The real work of building software that matters happens before you write a single line. And long after. It happens when you've felt a problem deeply enough that you understand what relief would actually look like. It happens when you've tried three approaches and watched people use them in ways you didn't expect. It happens when you delete something you spent weeks on because it was clever but not useful.
AI cannot iterate the way a human can because AI doesn't have lived experience. It doesn't have the scar tissue that comes from wrestling with a problem. It can generate the code for a solution, but it can't feel which solutions actually work.
You only know that by building slowly, by paying attention to how the software feels in your hands, by watching how people use it, by iterating from real feedback rather than hypothetical scenarios.
Here's what the feed misses: software is cheap, but understanding is still expensive.
Anyone can ask for code now. What they can't generate is a deep intuition about a problem space. That comes from living it, from making mistakes, from caring enough to keep showing up when the easy path is to just ship something that sort of works.
The software business isn't dead. It's just that the bar has moved. The skill of writing code is less valuable now. The skill of identifying a real problem, understanding it deeply, and building something that actually solves it is as valuable as it ever was.
If anything, it's more valuable. Because now there's more noise than ever. The people who can build something that actually matters will stand out.
I'm starting a solo software business despite the noise. Not because I think the market isn't competitive. But because I believe that the things worth building don't come from a single prompt. They come from curiosity, from iteration, from caring more about the problem than about the code.
Software is cheap. Meaning isn't.
The next time you see a post about how AI is killing the software business, remember: they're talking about code. They're not talking about understanding. And understanding is why people actually sign up.
The barriers have fallen. The code writes itself. The market is flooded.
But something's missing from this story.
The story assumes that software is code.
It assumes that the hard part of building a tool is writing the code to make it work. It assumes that once you can generate apps in minutes, what makes something worth building collapses.
I tried this. I sat down with a clear plan and a vision, and I used AI to generate what I asked for. Feature by feature, component by component. And the code worked. It was impressive.
But the tool didn't feel right. It felt hollow. Like a house with all the furniture in the right rooms but nobody lives there.
Because code is not software. Code is just the artifact.
The real work of building software that matters happens before you write a single line. And long after. It happens when you've felt a problem deeply enough that you understand what relief would actually look like. It happens when you've tried three approaches and watched people use them in ways you didn't expect. It happens when you delete something you spent weeks on because it was clever but not useful.
AI cannot iterate the way a human can because AI doesn't have lived experience. It doesn't have the scar tissue that comes from wrestling with a problem. It can generate the code for a solution, but it can't feel which solutions actually work.
You only know that by building slowly, by paying attention to how the software feels in your hands, by watching how people use it, by iterating from real feedback rather than hypothetical scenarios.
Here's what the feed misses: software is cheap, but understanding is still expensive.
Anyone can ask for code now. What they can't generate is a deep intuition about a problem space. That comes from living it, from making mistakes, from caring enough to keep showing up when the easy path is to just ship something that sort of works.
The software business isn't dead. It's just that the bar has moved. The skill of writing code is less valuable now. The skill of identifying a real problem, understanding it deeply, and building something that actually solves it is as valuable as it ever was.
If anything, it's more valuable. Because now there's more noise than ever. The people who can build something that actually matters will stand out.
I'm starting a solo software business despite the noise. Not because I think the market isn't competitive. But because I believe that the things worth building don't come from a single prompt. They come from curiosity, from iteration, from caring more about the problem than about the code.
Software is cheap. Meaning isn't.
The next time you see a post about how AI is killing the software business, remember: they're talking about code. They're not talking about understanding. And understanding is why people actually sign up.