Penya: a year, a weekend, and an audience of (probably) one

Let's be clear: I haven't been diagnosed with ADHD, but there are signs! Somewhere on this site I've already confessed to the 15+ GitHub repos I'm sitting on, most of them half-baked projects, aside from the one that spawned this very website, and one other I'll keep top secret for now. Well, last week, I actually finished another one. The idea had been swirling around my head for a year, I'd made a couple of attempts at it, and it's finally come to life.

It started with the track SING|DANCE|LOVE|GLOW. The final version was ready on 2nd June 2025, but I didn't publish it until May 2026 — partly because I kept thinking I could squeeze more joy out of that song than its sound and rhythm already gave me. And so Glowspiration was born.

My first instinct was an Android app. I downloaded Android Studio, watched the tutorials, and built something with confetti, streaks and a hundred-odd daily prompts. It was, in fact, the very first GitHub repo I ever created. The start of the pile, if you like. And it worked! But it looked clunky, it felt clunky, and somewhere along the way I realised that getting an app onto the Play Store isn't exactly free or instant. The flame went out. One more for the half-baked collection.

So I tried again, this time on the web. I was very new to what people now call "vibe coding". I did my best in VS Code, and the result was rough enough that I never made it public. I hid it behind a login on a quiet subdomain and told myself I'd come back to it. That second attempt is, funnily enough, where the name Penya was actually born. Then life did its thing, and I forgot all about it for almost a year. (The project is still sitting in my Firebase account, now politely marked for deletion.)

Here's the part I didn't expect. When I came back to the idea 2 weekends ago, the tools hadn't really changed. I had. Since the first two attempts I had somehow turned into someone far more fluent with all of this, and the rebuild (now live at penya.etapuwa.com) came together at a pace that really surprised me.

What changed wasn't that I typed faster. It was that I stopped trying to be the one with all the answers. I started by asking Claude Code to read my old Glowspiration concept and boil it down into a brief. And then, before it wrote a single line, I asked it to interrogate me. Which colours? How many pages? What lives on each one? What happens the moment you've done today's prompt? Some of those were things I'd never actually decided, but I definitely had been carrying a blurry picture around in my head about how I wanted it to look. The questions dragged the real answers out of me. When I read the spec back, I added the handful of things it still hadn't thought to ask, then handed the whole brief to Claude design and told it to keep questioning me too. About half an hour later I had a look I genuinely liked. Not because the AI had guessed well, but because it had made me say out loud what I wanted.

Then came the fiddly but fun part: tweaking. It was here that I decided I wanted to add a few affiliate links, and I asked specifically for a graceful, non-intrusive way to do it. So there's a quiet corner called Finds. Four picks, one per theme, with a plain disclosure and zero interruption to the ritual. If anyone ever buys through one, it helps keep Penya free. If nobody does, no harm done. I'd just like people to understand why it's there.

And then I handed the whole thing back to Claude Code and let it build. If you could see the commit history, you'd see the pace for yourself: most of it across a single weekend, then a few bits and pieces done during the week.

The real lesson, the one I keep turning over, is this: a year ago the bottleneck was never Android Studio, or VS Code, or any tool at all. It was me, treating these things like a vending machine: feed in the perfect prompt, collect the perfect app from the tray below. And when it didn't work, I assumed I'd just worded the prompt badly. What I'd actually got wrong was my mindset. The key, at least for me, turned out to be almost embarrassingly simple: stop describing what I want, and start letting the AI ask me what I want. The clarifying questions did the heavy lifting twice over — they built the spec, and they forced me to finally decide what the thing even was. A vague, year-old wish became a finished product in a single weekend. Plus, for once, a repo that isn't half-baked.

So why pour all this into something that might only ever have an audience of one? Because it's fun. Because I'm learning things that I think will help my career. And even if they don't, I just love learning. I'm curious about basically everything, sometimes to my own detriment. But mostly because, on the off chance it reaches one other person and gives them a single brighter moment in their day, that would make me ridiculously happy.

So that's Penya. Sing, dance, love and glow. Maybe just for me, and maybe for you too.

And bonus: as I was writing this, I deleted the Glowspiration repo. That clunky first attempt is what got me here, but I didn't need to keep it to remember that. Some things you carry forward; some you just let go. Half-baked was exactly the right place to start. Let's see what other projects I can "complete" or let go of in the coming weeks and months!