I Released Deskspill — A Small Arcade Game Developed with AI Coding Tools
I recently released Deskspill, a free Windows arcade game developed with the help of AI tools.
I'm not going to pretend I wrote all the code myself. In fact, one of the goals of this project was to see how far I could take a small game idea by focusing on game design, testing, and iteration while relying on AI for much of the technical implementation.
The result is Deskspill, an endless arcade game about surviving a growing pile of office paperwork.
You can play it here: https://waffle913.itch.io/deskspill
The idea behind Deskspill
The concept is pretty simple.
Imagine sitting at an office desk where sticky notes constantly appear, each containing a small instruction.
Some ask you to press a key. Others require typing a word, completing a sequence, holding a key, or performing a mouse action.
You can select and drag the notes around your desk to organize your workspace.
At first, everything is manageable.
Then the notes start appearing faster.
You're halfway through typing a word when another task arrives. You lose your concentration, make a mistake, and suddenly the desk starts filling up.
Let 10 unfinished notes accumulate, and it's Game Over.
The difficulty doesn't come from complicated individual tasks. It comes from trying to complete simple actions while constantly being interrupted.
How AI was used to develop the game
I want to be transparent about the development process.
I didn't manually program Deskspill from scratch.
My role was primarily to define the concept, decide how the mechanics should work, test playable versions, identify problems, and determine what needed to change.
I used ChatGPT to help turn gameplay ideas and feedback into more structured technical instructions.
Codex handled the coding and implementation in Godot, based on those requirements and subsequent iterations.
AI-generated artwork was also used for parts of the game's visual identity, including its promotional artwork and office environment.
The music and sound effects were synthesized through a Python script created by Codex rather than taken from an existing commercial soundtrack.
This wasn't a process where I entered a single prompt and received a finished game.
It involved multiple playable versions, testing, adjustments, and occasionally abandoning features that seemed promising on paper.
One feature that didn't survive testing
A good example was the round-based gameplay system.
At one point, I experimented with dividing the game into increasingly difficult rounds.
The idea was to give players short breaks between periods of intense multitasking.
It sounded reasonable.
But after testing it, I realized the rounds were actually making the game less interesting.
The breaks interrupted the accumulation of tasks, which was precisely what made the original concept enjoyable.
So I removed the round system and returned to continuous Endless gameplay.
That experience reinforced something important for me:
A feature that sounds good in a design document doesn't necessarily make the game better.
Actually playing the game matters.
What's included in Deskspill?
The released version features:
- Endless gameplay with increasing difficulty.
- Several types of keyboard and mouse challenges.
- Draggable sticky notes.
- A corporate office-themed workspace.
- Music and sound effects.
- Scoring and streak rewards.
- Local high scores.
- Offline Windows gameplay.
There are no accounts or online services required.
The goal is simply to survive as long as possible and beat your previous score.
What I learned from the process
One of the biggest takeaways was that AI-assisted development still requires decisions.
An AI can implement a feature, but that doesn't automatically make the feature enjoyable.
Someone still needs to test the game, recognize when something feels wrong, decide what should change, and keep the overall experience coherent.
The project also made it possible for me to experiment with game design without needing to personally implement every technical detail.
That doesn't replace the work of experienced programmers or game developers. It does, however, make small experiments like Deskspill much more accessible.
For me, that was the most interesting part of the project.
Deskspill is available now!
Deskspill is free to download and play on Windows, with optional donations for anyone who wants to support its development.
I'm also interested in feedback from both players and developers.
Does the game get difficult too quickly? Are the instructions intuitive? Does the experience make you want to try again after losing?
And if you've experimented with AI-assisted game development yourself, I'd be interested in hearing about your experience.
Play Deskspill on itch.io:
https://waffle913.itch.io/deskspill
Thanks for checking it out!
Everything is urgent. Especially the things that were due yesterday.
Top comments (0)