When Vibe Coding Creates More Clients

We’re now about 3 years into being 6 months from software development jobs being eliminated. If you work in the space, I get it, it’s exhausting. After all, why would anyone pay premium developer salaries when they can just describe what they want and watch the code magically appear?

Well, except for the copyright issues. And the security issues. And the privacy issues. And the risk of producing something they can’t maintain. And the token costs are going up. And the fact that a non-trivial number of consumers hate everything about AI.

But, I digress. In this post, I want to offer a counter-data point. It’s only one, and it’s anecdotal, so I own that up front. But it comes from someone with skin in the game, and connects to an economic pattern that I’ve mentioned previously.

Meet Mr. Dev X

He and I go back a long way. He’s a talented developer, and while I pivoted into education services and technical consulting, he pivoted into owning a developer agency. He’s definitely more pro-AI than I am, but only because he has the skills required to harness it responsibly. He’s not a doomer or a hype merchant. Since we both own businesses, we often talk shop.

Lately, he’s been getting a significant amount of work that he hasn’t had before.

The New Type of Client

My friend has closed several new clients over the last few months, and they’re all clients who, until very recently, would never have considered a dev agency at all. Not because they didn’t have ideas, but because commissioning a custom software project from scratch was not a thing they were going to spend money on.

Developer time isn’t cheap, and when you’re a non-technical aspirant, taking on the risks not only of designing something from scratch, but also not getting ripped off by the numerous shady players and low-quality offshore teams, simply doesn’t align with the risk tolerance.

But in a recent conversation, my friend shared this trend. Something has changed, and I think it will benefit skilled professionals. These people now have access to vibe coding tools and access to AI-assisted development environments. And, as much as I crap on those who toss up some half-assed, insecure slop that doesn’t work out, there is a subset of people who are genuinely prototyping and acutely aware that they don’t have the skills to produce production-grade apps.

From my friend's description of the initial conversations, a common theme emerges: performance and security issues, no proper hosting setup, no backups, and no architecture conducive to maintainability.

And so, they show up at my friend’s door. Not to commission a project from scratch, but to finish something. And the sales cycle is relatively quick and easy because they’re willing to pay. Now, I was surprised to hear this, but the more I thought about it, the more sense it made.

These new clients are exhibiting sunk-cost logic.

They’ve already invested time, energy, and some money in their project. My hypothesis is that they can’t stomach walking away. Their prototype feels so close; if they could just iron out those “little details”, they’d have a real shot at making some money in the market or improving their existing business with custom tools.

I’ve been telling would-be entrepreneurs for years that Apps are a much riskier investment than more traditional businesses, but some people do make money, sometimes.

So, my friend is getting more work as a “slop janitor”. His team audits the code, refactors it into something maintainable, sets up the infrastructure, security, and hosting… You know, the parts that aren’t easy, fun, or interesting to the average vibe coder.

One new client with these needs is a data point; closing a handful in a month with absolutely no marketing effort around it is a trend. And none of these clients, absolutely zero of them, would have ever bought a project from scratch.

Enter Jevons Paradox

In 1865, William Stanley Jevons noticed something counterintuitive about coal: as steam engines got more efficient and coal became cheaper to use, consumption of coal went up, not down. More efficiency meant more economic activity, and more economic activity meant more coal.

That observation became Jevons Paradox: when a resource becomes cheaper or easier to use, the consumption tends to increase.

Now apply that to software development. If AI tools are making building prototypes dramatically cheaper and more accessible, that lowers the perceived risk of starting a software project.

So a new class of would-be builders is entering the market, people who previously self-selected out of the "commission a software project" category entirely. And when they hit the limits of what vibe coding can do, they only have two choices: give up or turn to a professional to get things production-ready.

The Replacement Narrative

We’re certainly not seeing much evidence that AI tools are increasing ROI, GDP, or productivity at scale. We’re also not seeing evidence of the replacement narrative. Most companies conducting layoffs and blaming AI show more evidence of a weak economy forcing cost cuts, not of shedding workers because AI can truly replace them.

Also, as a business owner, I can’t imagine why any “leader” who found even a 20% efficiency gain with AI wouldn’t immediately put those gains to work on better products and services; cutting to maintain the status quo means competitors who don’t do that will pass you. It’s the mark of a very stupid leader.

The hype says AI eliminates the need for expertise. Jevons suggests it may grow the pool of people who eventually discover they need it.

I’ll keep watching my friend’s business with interest, but I wonder, are any of you freelancers out there seeing something similar?