How I Replaced a $99/Month SaaS With an R800 Vibe-Coded App

I was paying $99/R2,000 a month for a tool that sends people a survey and emails them a PDF.

That's it. That's what it did. Survey. Score. PDF. Two thousand rand. Every month. For three months.

And then one Tuesday, I opened Bolt.new and mass-ruined my own week.

Score App is one of those tools that does exactly one thing really well — and charges you like it invented the concept. You build a scorecard, people fill it in, they get a report. Clean. Simple. Expensive.

I was using it to run my 5-Fit Business Scorecard — a 35-question assessment that tells founders which part of their business is on fire (and not in the good way). It worked. But every month when that R2,000 invoice landed, I'd stare at it and think: I'm paying premium rent for a glorified Google Form with a calculator.

The thing is, I couldn't customise it the way I wanted. I wanted cohort comparisons. I wanted to see trends across submissions. I wanted to control the email flow without navigating someone else's clunky dashboard. But Score App said no — or rather, Score App said "upgrade to Enterprise", which is corporate-speak for no.

So I did what any reasonable person with a Bolt.new subscription and too much confidence would do. I vibe coded the entire thing.

Here's how it went:

Step 1: I wrote a detailed spec in Notion — every screen, every flow, every edge case. This is the part most people skip. Don't skip it. Your AI coding tool is only as good as the brief you give it.

Step 2: I uploaded everything into Lovable first and burned through all the free credits building a working prototype. Then — and this is the move — I asked Lovable to rewrite the spec based on what it had built. That refined spec became my golden ticket.

Step 3: I took that spec into Bolt.new and rebuilt the whole thing with fresh credits. One day. One very long, very caffeinated day.

The result? An app that does more than Score App ever did:

Total cost: R400 + R400 in extra Bolt credits when something broke. Which brings me to...

Here's the part nobody tells you about vibe coding: it will break your things while trying to fix your things.

I'd built the front end first — the survey form, the quiz flow, the whole respondent experience — and it was working beautifully. Then I started building the admin dashboard. And while Bolt was building the management side, it quietly rewrote my survey form. The form I'd already finished. The form that was working.

Suddenly the survey was broken, but the backend was unfinished.

I tried to fix it. Ran out of credits. Had to buy more credits just to undo the damage, rebuild the survey, and then finish the backend I was trying to build in the first place. It was like paying a builder to renovate your kitchen and coming home to find they'd also knocked down a load-bearing wall in the lounge — helpfully, of course.

The lesson? Always tell your AI: "We are troubleshooting. You are not allowed to execute right now."

If you don't set that guardrail, vibe coding tools will happily recode your entire front end while you're asking a hypothetical question. They're eager like that. Dangerously eager.

This is the real question. And I think the answer is simpler than most people make it:

Vibe code when you need an internal tool that solves a current problem.

Don't vibe code your next big product launch. Don't use it to build something you're selling to thousands of people on day one. But if you're paying for a SaaS tool that does one specific thing, and you keep wishing it did that thing slightly differently — that's your sign.

The difference between "build" and "productise" matters. I built this scorecard app for me. It runs my assessment, collects my data, and feeds my admin dashboard. If it gets traction and people start using it at scale, then I'll invest in proper backend development. But for now? It works. It's mine. And it costs me nothing per month.

That's the real power of vibe coding for founders: you stop renting tools and start owning systems.

One of my clients vibe coded his own Circle community platform replacement and his own Luma replacement. He spent more on credits than I did, but compared to what he was paying monthly for both platforms? He'll break even in no time.

The tools aren't perfect. The process is messy. You will want to throw your laptop. But when you come out the other side with something that does exactly what you need — and you can fix it yourself when it breaks — there's no subscription invoice in the world that compares to that feeling.

Curious what your business actually looks like under the hood? I built this scorecard to find out.

→ Take the free 5-Fit Scorecard and see which part of your business needs attention — it takes about 12 minutes and you'll get a personalised report with specific next steps.

Yes, it's the vibe-coded one. Yes, it works. Probably.