Bubble Serenity
Developer diary... a developer? Me? :))
How AI, an ageing Lenovo and a little stubbornness turned an idea and good intentions into a real project getting off the ground.
Hi! My name is Shlomo, and I am developing Bubble Serenity — Relax Village Puzzle.
I should admit straight away that I am new to the games industry. Some time ago, I took courses in Unity and C# and learned the basics of programming, but I never got as far as making a game of my own. There was always a reason to put the idea off: not enough knowledge, tools that seemed too complicated, too much work for one person.
Then artificial intelligence came along. And my endless “someday” unexpectedly turned into “why not now?”
From Tower Defense to colorful bubbles
The story began in mid-July 2026. Although, strictly speaking, it began even earlier.
Originally, I dreamed of creating a 3D Tower Defense strategy game. And, as is often the case with beginners, it was not exactly a modest idea. The mechanics grew more complicated faster than I could work out how to implement them.
I was counting on help from 3D model generators such as Meshy and Tripo, but at the time, neither their capabilities nor my knowledge were enough for a full development project.
Eventually, I decided to be sensible and start with something simple. A 2D Bubble Shooter, for example.
Spoiler: the sensible part did not quite work out.
An ordinary game of shooting colorful bubbles gradually acquired Heroes, combat abilities, village development, resources, a workshop and competitive PvP battles.
Still, I have kept one important promise to myself so far: the main game has remained two-dimensional.
My little team of artificial intelligences
I am working on the game on my own, though I do not think the words “on my own” have ever had quite such an unusual meaning.
My main helpers are ChatGPT, Claude Code and Codex. Through MCP, I connected the development tools to the working project. I create sprites with GPT and experiment with animation using video in Grok, followed by further processing with Claude.
My role goes something like this: come up with an idea, explain it, choose, check, occasionally argue with the AI, then explain once again why the first version was actually better.
All of this happens on a rather elderly Lenovo computer, which probably never imagined it would spend its later years developing games.
AI has been an incredible helper. But I soon discovered that getting artificial intelligence to write code and creating a good game are two very different things.
First came the documents. Lots of documents
Before getting seriously into the code, I started describing the game in Markdown files.
One document covered the combat system, another the Heroes, and a third the economy. Then came architecture, balance, monetization, localization and the development roadmap.
I thought that if I planned everything carefully enough, development would follow the plan.
The first major refactoring came about a month later. The second followed another couple of weeks after that.
Since then, I have rewritten almost all of the project documentation several times. Mechanics changed, as did the visual style and the structure of individual systems.
That revealed something important about developing with AI: give it imprecise instructions, and it can build exactly what you do not actually need — very quickly and very convincingly.
So these days, my Markdown files are more than notes. They are the foundation of the entire project.
Two pieces of advice that changed a lot
My brother, who has been following my adventures in development, has given me plenty of useful advice. Two suggestions proved especially important.
The first was to stop trying to make the entire game at once and focus on a player's first five minutes. What will they see? How will they feel? Will they want to keep playing?
The second was to split work on the project between two repositories so that different areas of development could move forward faster.
Those suggestions helped me move from endless planning to more tangible results.
What is there today?
Less than four months have passed since mid-July.
Bubble Serenity already has its own visual world, combat system, Heroes, interfaces and a first version of the tutorial. It also has the website where you are reading these words. Most importantly, I have put together the first Android test build.
Of course, it is not a finished game yet. There is a huge amount of work ahead, especially on game balance, usability and that elusive feeling that makes you want to play “just one more time”.
But for the first time, I can do more than talk about my idea. I can run it on a phone. And that is a wonderful feeling.
What comes next?
In this diary, I will share development progress, show screenshots, and write about plans, good decisions and mistakes. Sometimes I will cover technical details; at other times, the doubts that inevitably come with trying something new.
I cannot be sure the game will be successful. But I very much want to finish it and create something people will truly enjoy playing.
I am also curious to see how far one person can go today with an idea, patience and a few very bright digital helpers.
With God's help, we keep going!
Shlomo
November 8, 2026
Publication dates in UTC