Devlog #01: The Gender-Reveal Game That Became a Studio

This is the first entry in the Shrines of Mythora devlog — an honest, build-in-public account of making my first game. No polish, no pretending I know what I'm doing. Just the real process, the wrong turns, and the moments it clicks.

To explain where this game is going, I have to start with where it came from.

It started as a gift

The whole thing began with a small exploration game I built for my wife — a gender reveal, wrapped in a few riddles and a final puzzle that, when solved, told her we were having another boy. That part of the story is personal and deserves its own telling, which I'll do separately in how it all started. For this devlog, what matters is what happened next.

A rough game that worked anyway

I'll be honest: the game wasn't good. The exploration was thin, and the final puzzle barely qualified as a puzzle. If you're being generous, it was held together with tape.

But it didn't matter. She loved it. The message landed. And that taught me something I keep coming back to: emotion beats polish. A clumsy thing that means something will out-perform a slick thing that means nothing, every single time.

From one-off gift to a product

My wife was the one who pushed it further. "Other people would love this," she kept saying. That nudge is the reason Four Hearts Studio exists at all. The idea wouldn't leave me alone — so I started planning what it would take to turn that first build (which, for the record, took me a few weeks, not the one night the story makes it sound like) into something I could actually give to strangers.

The shape of the game (and why it survived)

When I sat down to plan properly, I expected to redesign everything. Instead, the original structure held up better than the original execution did. It stayed almost exactly as it was:

• The player arrives and is given their task.
• They solve three puzzles, each one personalised to them.
• Once all three are done, a final puzzle unlocks — and the message is revealed.

Three acts. A clear loop. A payoff that's earned, not handed over. The bones were right the first time; everything since has been about giving them a body.

The feeling I'm chasing

What I want is hard to put in a spec. I want the player to feel like they're inside a story that's unfolding as they move through it — not reading it, playing it. A world that reacts to them. And then the part that makes it ours: I want it to know their name, their in-jokes, the moments only they would recognise.

That's the bar. A real-time story that feels like it was written for one person, because it was.

How personalisation actually works

This is the part I'm most excited to share, because it's where my background actually helps. Here's the pipeline I've designed to turn a buyer's memories into a playable story:

1. The memory-gathering agent

The buyer has a conversation with an AI agent whose only job is to draw out the good stuff — memories, jokes, shared moments, the small specific details that make a person them. It's conversational, not a form.

2. The story writer

Once the agent has enough, it summarises and hands off to an AI story-writer. This is where the quirks get woven into the narrative arc — not bolted on as easter eggs, but threaded through the story so it feels native.

3. Riddles and the final message

From there we generate the three riddles tied to the three puzzles, plus the final message — the whole reason the gift exists.

4. The voiceover

Then AI voiceover narrates it, giving the whole thing that read-aloud, storybook feel I keep chasing.

5. The game pulls it in

Finally, the game loads the custom story data at runtime. If it all works, the player walks into something that quietly knows them.

Web brain, game-dev wall

Here's the honest split. The backend, the AI agents, the data plumbing — that came easily. I've spent years in web development, so that half feels like home.

The game itself is another level entirely. Spatial design, level layout, making a 3D world feel alive — none of that is in my muscle memory, and I've hit walls I didn't know existed. What's kept me moving is the partnership: my logical, systems-brain paired with Claude as a thinking partner. It's not "AI builds my game." It's closer to having a collaborator who's good at exactly the things I'm not.

Next time

In Devlog #02, I hit the first of those walls hard: blocking out the island. I'd never done it, I had no idea where to start, and I froze for longer than I'd like to admit — until one perspective shift turned a blank scene into a walkable island in about an hour.

If you want to follow the build as it happens, this is the place. More soon.