Watch the Youtube Video

I got an opportunity to build custom software for a group of companies owned by friends. But I was biting off more than I could chew, and somewhere in the middle of it, I capitulated.

This wasn’t an app. It was becoming a full-fledged SaaS management system.

Eventually, I had to hand it over to the team at a company I co-founded a long time ago—a business that deals with IT infrastructure management, enterprise ERP systems, and vertical software.

My co-founder wanted nothing to do with apps or consumer products, and that is partly how I ended up on this strange quest of building apps by myself. At least, that’s part of the story.

Sadly, this project was taken out of my hands by knowledge.

Yes, knowledge.

The knowledge of recognizing when something is going to need a team, even in this AI era.

I’ve been thinking that, in some ways, vibe-coding providers—at least in their current form—should come with a giant disclaimer:

“We help you build a prototype.”

Because building something that works once is not the same as building software that survives real users, real data, changing requirements, and every strange edge case nobody thought about at the beginning.

Sadly, I’m not ignorant enough about software or business systems to pretend otherwise.

But from all of this, I was left with a piece of code.

So I decided to turn it into a small-business bookkeeping and invoicing app.

Simple enough, right?

That has been a refreshing reminder of why you probably shouldn’t build even a “small-scope” app with this kind of ambition.

That is where knowledge betrayed me again—or maybe I should say confidence did.

Because even a small bookkeeping app has invoices, payments, expenses, taxes, balances, receivables, payables, edits, deletions, PDFs, and all the ugly edge cases that appear when money is involved. A value changing in one place can quietly make another part of the system wrong.

And yet, I’m building it.

Apparently, it’s my new app now.

Even if it takes me a few more weeks and every AI token I can throw at it.

Dealing with each edge case is a nightmare.

But I guess that is the inheritance of saying no to the bigger thing: sometimes you walk away carrying a smaller piece of it with you.

And then, somehow, that smaller piece becomes the next thing you build.