The End of Software As We Know It
I keep hearing a lot of “proper” coders say that AI-generated code is untrustworthy. That AI hallucinates. That it writes bugs. That you still need a real developer to review everything.
Hogwash.
Yes, AI makes mistakes. So do humans. Yes, AI code needs reviews. So does humans. The difference is that AI is getting better at a pace that makes human improvement look glacial. The code Claude writes today would have been impossible six months ago. Six months from now? Another world entirely. The trajectory is, at least to me, is undeniable if you’re paying attention.
We are witnessing the end of software as we know it. Not the end of software — the end of how we make it, buy it, and think about it.
Agent Interfaces Will Be Table Stakes
Right now, most software is built for humans to click through. Menus, buttons, forms. But that’s already starting to feel antiquated. If you want your product to survive the next few years, you’re going to need an agent interface — a way for AI to interact with your service programmatically, conversationally, intelligently.
APIs were the first step. Agent interfaces are the next. The companies that don’t offer them will be routed around by the ones that do. Your competitor won’t just have a better UI — they’ll have an AI that can use their product on behalf of customers. That’s a different game entirely.
Morphable Apps Will Dominate
Think about how many apps exist to solve narrow, specific workflows. Project management tools, CRM systems, invoicing apps, scheduling tools — thousands of them, each with their own opinions about how you should work.
What happens when an AI can morph an app to fit your exact needs on the fly? You don’t need 50 specialized tools when one intelligent tool can reshape itself based on what you’re trying to accomplish. The long tail of single-purpose apps is going to collapse. Not disappear entirely — but shrink dramatically.
The winners will be platforms that embrace this malleability. The losers will be the ones clinging to rigid feature sets.
Entertainment Survives (For Now)
Games, movies, music — the creative, experiential stuff — will be the last to fall. Not because AI can’t generate entertainment (it can, and will), but because the experience is the product. People don’t play a game to get a task done. They play to feel something.
That’s harder to automate away. Harder, but not impossible. I’m watching this space closely.
Software Gets Cheap
When anyone can spin up an app in minutes instead of months, what happens to the price of software? It falls. Hard.
We’re entering an era of commodity software. The value won’t be in the code — it will be in the data, the integrations, the trust, the brand. Writing software will be like writing a document: everyone can do it, so doing it isn’t special anymore. What you do with it is what matters.
The Critics Are Looking Backward
Every time I hear someone dismiss AI-generated code, I think: you’re judging tomorrow’s technology by yesterday’s limitations. Yes, there were hallucinations. Yes, there were bugs. But have you used it lately? Have you seen the improvement curve?
I’m not saying we’re there yet. I’m saying we’re getting there faster than most people realize. And by the time the skeptics come around, the world will have already moved on without them.
The end of software as we know it isn’t a threat. It’s exciting!