I Built an App Without Writing a Single Line of Code — Here's What Actually Happened

I want to be upfront about something: I am not a developer. I've never written a line of code in my life, at least not intentionally. The closest I'd ever come to "building software" was setting up a spreadsheet with some fancy formulas and feeling unreasonably proud of myself during my time at BP.

So when someone in my circle mentioned Replit's AI-powered app builder, I was curious but skeptical. Sure, I thought. Another tool that promises to make things simple but quietly assumes you already know what you're doing.

Two weeks ago, I decided to just try it. What followed genuinely surprised me.

The Idea: A Problem I'd Been Ignoring for Years

The thing is, I already had something in mind I wanted to build, nothing fancy, not unique. Anyone who's ever gone on a group trip, shared a flat, or just grabbed lunch with colleagues regularly knows the pain — someone pays, someone forgets to transfer their share, someone's mental math is slightly off, and suddenly there's this awkward cloud hanging over what should've been a perfectly pleasant experience.

I know there are many tools that already address the issue, but still I wanted a simple spend split tool. Just something where a group of friends or colleagues could log shared expenses, see who owes what, and actually keep track of it over time without relying on a running thread of messages or a notes app that only one person can see.

It wasn't a groundbreaking idea. But it was *my* idea, and I wanted to build it myself.

The First Few Minutes Were... Surprisingly Calm

I expected to feel lost immediately. Instead, the experience felt more like describing an idea to a colleague than wrestling with any kind of technical interface. I typed out what I wanted — a group expense tracker where you could add members, log costs, assign who paid, and automatically calculate each person's share — and Replit's AI just... started building.

Watching it work in real time was the strangest thing. Code was appearing on one side of the screen, and on the other, an actual working interface was taking shape. It wasn't a mockup. It wasn't a template with placeholder text. It was a functional spend-splitting app being assembled in front of my eyes, based on nothing more than what I'd described in plain English.

I genuinely didn't know what to do with myself. I just sat there for a moment.

Then Came the Errors (Of Course)

Look, it wasn't all smooth sailing. I'd be lying if I said the whole thing just worked perfectly from the jump.

At a couple of points, the app hit snags. The expense calculations weren't splitting correctly when the group size changed. A feature I'd asked for — being able to mark a debt as "settled" — wasn't behaving right. There were moments where the interface looked off, or a button did something completely different from what I'd intended.

But here's the thing I didn't expect: the fix wasn't technical. It was communicative.

The errors weren't problems I needed to debug in code — they were misunderstandings between what I had described and what the AI had interpreted. When I said "split equally," I hadn't specified what should happen when someone opts out of a particular expense. When I asked for a running balance, I hadn't explained that I wanted it broken down per person, not just as a group total.

When I went back and gave more specific instructions — more context, clearer boundaries, a better explanation of what I actually meant — the results improved. Every single time.

It felt less like troubleshooting software and more like refining a brief. And that's a skill I actually have.

Less Than an Hour Later

I'm not exaggerating when I say the whole thing came together in under an hour. A working spend split tool — one where you can add your group, log any expense, see an instant breakdown of who owes what, and track settlements over time. Built by me, someone with zero development background, in a single sitting.

Is it competing with apps like Splitwise? No. But it works. It does exactly what I needed it to do.

That's the part that keeps sticking with me. Not just the convenience of building it, but what it means. The gap between "I wish there was a tool that did this" and "here is the tool that does this" used to require either technical skills I don't have, or the time and money to bring in someone who does. Now, for a certain category of problem, that gap is just... a well-worded description.

What This Actually Changes

I'm not here to claim that tools like Replit replace developers or that anyone can build anything with no expertise. That's not what I experienced, and I don't think that's the honest takeaway.

What I *can* say is that my relationship to the phrase "I wish there was an app for this" has fundamentally changed. I have a list of small, specific tools I've always wanted — things too niche to exist in the App Store, too personal to outsource. That list looks very different to me now — and it's worth noting that AI-powered tools are what made it possible, not just effort or time.

For someone like me — an ideas person, not a technical person — that's a bigger shift than it might sound.

I'll definitely be going back.

What This Means for Us in Procurement

For those of us in procurement specifically, I don't envision this replacing any of the platforms and systems we already rely on. But I can absolutely see colleagues starting to build small procurement automation tools that fill the gaps those systems never quite reach — something like:

The kind of thing that usually lives in a spreadsheet forever because building something better always felt out of reach.

What strikes me is that this sits at the edge of something bigger. The conversation around AI in procurement has largely focused on enterprise platforms — Coupa, SAP Ariba, Jaggaer — and rightly so. But there's a quieter layer of procurement transformation happening at the team level, where practitioners are starting to close the gap between "I wish this existed" and "I built it this afternoon." That's not the same as agentic procurement workflows or large-scale procurement automation — but it's directionally the same instinct. And for anyone thinking about where AI fits in their day-to-day work, that instinct is worth following.