Your App Won’t Live Forever
The revolution is bringing home some truisms that we’ve known about for eons. THAT revolution. The AI one, you know? Yes, I knew that you’d be aware.
The fact that some kid in a basement is able to write an app from prompts alone is bringing home the real cultural change. When you ask a question and compiled code is returned inevitably you’re generating tiny, throwaway apps.
These are temporary, like a feather in the wind. By definition.
Code Got Cheap
The vibe coding revolution isn’t helping anyone to write the new Salesforce. We aren’t seeing the next generation of software that will solve the really hard problems humanity faces. Rather, we’re seeing tiny throwaway apps.
We’ve entered a new scratchpad era where you can build software, use it and then throw it away without looking back.
This software (vibe coded or not) is a personal utility belt. Code that is very different from professional software that would be used for a whole team, and requires onboarding documentation. Once and done is very different from software which is considered, generic and maintainable.
Remember when people would fire up Excel to solve a one-off problem? Today that’s simply Claude code.
Cheap Code Produces Expensive Problems
The fact is AI tools have gotten frighteningly good. The CLI-first movement is back. Developers, makers, and even non-tech folks are jumping back into the terminal because it gives them control. Not abstraction. Not a pretty GUI with limited options. Real control.
Code is cheap now, but software isn’t. It never was, and that isn’t going to change soon.
A five-minute CRUD app is all very well (and lots of fun) until the DOM changes. The API fails or the input CSV format changes. Finding where the error is becomes a tricky (at best) problem that needs to be solved, and typically that work is completed by a competent engineer.
The price of good code judgement just increased inversely to the price of producing the code itself.
This was always the case though. Half-baked ideas that break at the first sign of an edge-case are cheap but well written software battle-hardened in production is a valuable commodity.
Build it in a weekend
If your social feed is anything like mine, you’re seeing developers brag about “$10K MRR” from an AI-built app they launched last Friday.
Sure, maybe some of them are real. But most are marketing stunts wearing a technical trench coat.
Because once you remove engineering as the differentiator, what’s left? Distribution. Taste. Timing. Most of these so-called “overnight successes” were launched into an audience that already cared. Not because the software is great, but because the audience is real.
The rest of us? We’ll launch into a void that won’t even pay for our time thinking about the idea.
The Limit of AI
This new generation of tools is magical. They strip away boilerplate, scaffold UIs, integrate APIs. They are brilliant assistants.
But that’s all they are.
They still make mistakes. They hallucinate APIs. They assume too much. You cannot skip the code review just because Claude wrote it. If anything, you should review AI code more carefully. Treat it like a PR from a junior dev who skipped lunch.
No. Your non-technical exec can’t just fire the engineering team now. Because these tools still need someone to know when things will break, and more importantly, why.
That’s not promptable. Your engineering manager isn’t going to be able to work this one out quite so easily on their own.
Who Thrives in This Mess?
The obvious answer is those who are peddling AI, that is Sam Altman and of course those providing the power to data centres.
Yet there are quiet winners in this race. Those who aren’t quite running chasing unicorns.
Experts solving their own repetitive pain points
Internal devs hacking together one-off tools for real teams
Power users who got tired of waiting for IT to approve their Zapier flow
Engineers who use AI to move faster, not skip thinking
Code might be free, but everything that matters is still costly. Good design, good decisions, and actual experience.
That won’t drop to a zero-dollar value anytime soon.
About The Author
Professional Software Developer “The Secret Developer” can be found on Twitter @TheSDeveloper and regularly publishes articles through Medium.com
The Secret Developer once built an app to track their productivity. It counted their PRs.