How I Vibe Coded WagTracker in an Afternoon (And Why It Started on a Beach)

It started with a walk on the beach.

I was chatting to my dog walker — the person who comes to the house, picks up the dogs, does the walk, brings them back. She mentioned offhand that the admin side of her job was a nightmare. Booking requests coming through WhatsApp. Invoices done by hand. No way to see who'd paid and who hadn't without digging through her messages.

That conversation stuck with me. And a few days later, I sat down and built WagTracker.

▶ Watch the full build story on YouTube

What is vibe coding?

Vibe coding is a way of building apps where you describe what you want in plain English and an AI writes the code. You're not typing every line yourself — you're having a conversation with the AI, guiding it toward the thing you want to exist.

Replit makes this work really well. You open it up, describe your project, and it starts building. You look at what it's done, tell it what to change, and gradually shape it into a working app.

It's not magic. You still need to understand what you're building and why. But you don't need to memorise how to set up a database before you can start. The barrier to getting something real on screen is much, much lower.

Start with the problem, not the code

One thing I always try to do before building anything: understand the problem from the person who has it.

My dog walker's problems were specific:

• Clients messaged her through WhatsApp and she had to manually note everything down somewhere
• She had no easy way to see her whole week at a glance
• Invoicing was a separate thing she did at the end of the month — and she often forgot who owed what

WagTracker needed to handle three things: bookings, client records, and payments. That was the brief.

What an afternoon of vibe coding actually looks like

Here's roughly how the build went:

Start with the shape. I described the app to the AI: a booking system for a solo dog walker. It set up the basic structure — a way to add clients, log walks, and mark them as paid or unpaid.

React to what it builds. The first version never looks like the final version. The AI makes choices — sometimes good ones, sometimes not. You look at what it produced and say "yes, keep that" or "no, change this bit."

Add the specific details. My dog walker wanted to see her week as a simple list, not a calendar grid. Small thing, but it matters. I described it. The AI adjusted it.

Test as you go. You click around, find things that don't work, describe the fix. This is where your brain matters most — you know what the app is supposed to do, and you can spot when it doesn't.

By the end of the afternoon, I had a working prototype. Not polished. But real.

Why this matters for learning to code

You might think vibe coding means you're not really learning. I'd push back on that.

When you vibe code, you make decisions constantly. What does the app need? What should happen when a user clicks this? Why is this bit behaving strangely? Those are programming questions — you're just not the one typing the answers yet.

Over time, you start reading the code the AI produces. You start understanding it. You start having opinions about it. And at some point, you can write it yourself.

Vibe coding isn't a shortcut around learning. It's a way in.

Watch the build

The YouTube video walks through the whole thing — the beach conversation, the Replit session, and the decisions along the way.

▶ Watch: I Built This App Because My Dog Walker Needed It

If you want to try WagTracker, it's a live demo at cognitocoding.com — click through and see the build for yourself.