Vibe coding, veggie gardens, and the build vs buy cycle

LinkedIn has been full of it lately. Vibe coding. AI-generated apps. The death of SaaS. The posts have been coming in waves — some triumphant, some panicked, most a bit of both.

The core claim goes something like this: now that anyone can build a working app in an afternoon using AI, why would a company pay thousands of dollars a month for software that does 80% of what they need and nothing else? Just build exactly what you need. Spin it up in a weekend. Cancel the subscriptions.

It’s a reasonable argument. And the technology is real — I’ve seen people build things that would have taken a small team weeks, in hours. The capability shift is genuine. But the conversation keeps missing something important, and I’ve been saying it in the comments of various LinkedIn posts enough times that it’s worth writing down properly.

This has happened before

Before SaaS existed, most companies built their own software.

Not with AI — with developers. Internal teams, outsourced agencies, consultants who would come in, build something bespoke, and hand it over. Every company that was serious about technology had a suite of custom tools: CRM systems, billing platforms, reporting dashboards, HR tools. If you needed it and couldn’t buy it off the shelf, you built it.

And for a while, that worked. The software did exactly what the business needed because it was built for exactly the business. No paying for features you’d never use. No compromising on a workflow that didn’t quite fit.

Then the maintenance started.

Bugs. Feature requests. The person who built it left. The technology it was built on aged. The business changed and the software didn’t. What started as a competitive asset slowly became a liability — a system that was technically functional but was absorbing engineering time that the company didn’t have and couldn’t justify spending on something that wasn’t the core product.

So they bought software instead.

Not because buying was cheaper in the short term. Often it wasn’t. But because the total cost of owning software that isn’t your core business is higher than the sticker price suggests. The companies that lived and breathed that problem space — CRM, billing, HR — could build and maintain it better, cheaper, and with less distraction. The build-it-yourself crowd eventually reached the same conclusion, one by one.

This cycle is not new. Vibe coding hasn’t changed the economics. It’s just made the first part of the cycle dramatically faster.

The veggie garden problem

I’ve been using an analogy in the comments of a few of these posts and it seems to resonate: the veggie garden.

A lot of people plant a vegetable garden at some point in their life. The motivation is clear — fresh food, lower costs, satisfaction of growing something yourself. And the start is genuinely great. You dig the bed, plant the seeds, water it for the first few weeks. It works.

How many of those gardens are still producing three years later?

In most cases, the answer is: not many. Because maintaining a garden year after year requires sustained attention and genuine interest in gardening. If your real passion is cooking, or your family, or your job — the garden gets less attention over time. Weeds come. Seasons pass. Eventually it’s just a patch of dirt.

Vibe coding apps have the same dynamic. The first version is exciting. You built it yourself, it does exactly what you need, and the technology made it genuinely easy. But software is never finished. Requirements change. Bugs surface. Someone on the team wants a feature the original version doesn’t have. The AI tools get better and the version you built six months ago starts to feel dated.

If your core skill and passion is the thing your business actually does — not building software — the probability that you’ll keep investing in the app drops off quickly. Not because you’re lazy, but because you have better things to do with your attention.

Where this leaves SaaS

I don’t think SaaS is dead. But I think some of it should be worried.

The category most at risk is point solutions. Tools that do one narrow thing. Lightweight CRMs, simple project trackers, basic invoicing tools, single-purpose workflow apps. The things where the core functionality is simple enough that a motivated person could genuinely replicate 90% of it in a weekend with AI. If your product competes on “it does the thing and the UX is nice,” that’s a real problem. The bar for building a functional version of “the thing” just dropped dramatically.

The category that’s less at risk — and this is where I’d put Bunny — is software where the complexity is real, structural, and domain-specific. Billing infrastructure. Revenue operations. CPQ. The kind of software where the surface looks manageable but the edge cases are everywhere, the integrations are deep, and the institutional knowledge of how to handle a mid-cycle subscription amendment or a multi-currency ramp deal took years to build.

You could vibe-code a billing system. People are going to try. And for simple use cases, it will work fine. But the moment the complexity kicks in — and in billing, it always does — you’re not maintaining a veggie garden anymore. You’re maintaining a farm. That requires a different kind of commitment.

The pattern repeats

The speed has changed. The pattern hasn’t.

Companies will build more of their own tooling over the next few years. Some of it will stick. Genuinely simple tools, built for specific workflows, maintained by people who care about them — those will survive and maybe thrive.

But a lot of it will follow the same arc that custom software always follows. The initial version works. The maintenance compounds. The business grows and the custom tool doesn’t keep up. The people who built it have moved on. Someone does the math on what it’s actually costing in engineering time and distraction, and the conclusion is the same one companies reached before SaaS existed: buy from someone whose whole job is to solve this problem well.

Vibe coding is real and it’s remarkable. It’s going to reshape a lot of things. But the economics of focus — doing what you’re actually good at and buying the rest from people who are — haven’t changed. They never do.