The story behind the island

I thought I'd add a bit here about me.

This game is a birthday gift to the lovely Sophie. I'm the creator of it — every detail, system, quest and little in-joke carefully crafted to make sure the game flows and feels balanced.

I didn't write a single line of code for it, though.

I've seen a lot online lately about the rise of AI, and the arguments against slop — the worry that it's a detriment to artists. In my own small way, I wanted to set out my stall that this isn't always the case. Yes, I had the vision for this, but not the technical know-how. It's only the second game I've ever made, and I've never taken on a project that grew as many legs as this one.

I built the whole thing using Anthropic's Claude Fable 5 and Opus 4.8 models, with the intro video and loading screen coming from Gemini Flash 2.5. I want this to be a little tale of how AI can be used as a tool to make something good — because without these tools, it simply wouldn't exist.

The music

This game threw up plenty of challenges. The first real one was the music. I knew I wanted to build a synth engine and generate the music, rather than making tracks in Suno (which I did toy with). I wanted it to feel relaxing and loop without getting annoying.

Time and again I'd set the key and the BPM, let it away with arpeggios… and it just wasn't right. In the end I wrote a little melody line myself, then told the synth engine which notes to play and for how long, which waveforms to use, and how many to layer — a challenge in itself. There were so many iterations before it sounded good to my ear. It's all subjective, of course, so if you hate it — sorry!

Pacing & economy

Pacing and economy balancing were two areas that demanded a lot of attention. Making sure the game flowed right, putting gates on the quests so the player doesn't just fly through the main story, and making sure it never felt like too much of a grind — all of that took a lot of back-and-forth, tweaking values.

My own worst enemy

Then there's me. I'm my own worst enemy. When I set out, I had a build order: map the island, then the characters, then all the mechanics — the brewing, the festivals, the shop — then the quests, then the polish, then the build (I've probably skipped a few steps there, but who's counting), and finally a full playtest order and checklist.

But my inability to focus meant I'd get partway through a playthrough and have ideas for entirely new mechanics — the garden plots and greenhouse, a complete overhaul of the object placement system, the final three acts! If I'd stuck to my original plan the game would have been done in half the time — but it wouldn't have been anywhere near as good. My favourite bit? Walking out of the brewery at dusk and seeing the light on Sophie's face. Chef's kiss. The lighting and graphical improvements were another rabbit hole that took far longer than I ever intended.

I'm super happy with how this turned out. It definitely won't be my last foray into making games with AI.