Welcome to Game Design Bites - a personal take on game design, straight from years of experience in the industry!
If this interests you, subscribe to get game design insights straight to your inbox for free!
[
](https://substackcdn.com/image/fetch/$s_!gc3C!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa4851123-e6cf-460d-84ac-f59f1f37b0a5_1080x810.jpeg)
Photo by Sebastien Bonneval on Unsplash
I have a graveyard of ideas that never made it past rough notes in my notes app. I always thought I’d do something with them eventually, but never quite got there.
Not because the designs were bad. Because there’s a big difference between “I know exactly what this should feel like” and “I can show you what this feels like” (especially when you don’t code). Every prototype attempt would hit the same wall: a blank project, confusing documentation, and a vision that would take months to reach something testable. So I did the easy thing and gave up on most of them.
I started looking for a workaround. My earlier post on "vibe coding" was the first real experiment — one game, an AI, and a workflow that turned out to be more about communication than code. The clearer my intent, the better the output.
I’ve continued to experiment with new ideas using this workflow. One of those experiments has resulted in my itch.io page: Playable Things.
🧭 Exploring AI as a Designer
It’s hard to ignore the pace of AI development. I’ve been hearing about new models and tools everywhere lately, and I wanted to read and try a bit more of it myself—not as an engineer, but as a designer looking for better ways to express my ideas.
I’ve heard the names like Claude Code, GPT Codex, Cursor, but I hadn’t really used them. The setup I eventually landed on was Google Antigravity and Google Stitch , mostly because it was free and approachable. I could install it and start playing around without a subscription gating every experiment. It felt less like fighting a tool and more like working with an incredibly fast, if very literal, collaborator.
The process is fairly straightforward: when I have a mechanic I want to test, I use Antigravity to build it out using the tools it provides. Once it’s in a playable state, I host it on itch.io to see how it feels.
Here are the first few prototypes I’ve managed to publish.
🕹️ YAWG: Definitions Meet the Grid
The latest and perhaps most polished experiment in this series is YAWG (short for Yet Another Word Game).
The design of YAWG focuses on a straightforward loop: you’re given a list of dictionary definitions and a shared grid of tiles, and your job is to fill in the blanks. If you want to try different combinations for a word, you can do so freely. The only constraint is the timer in Arcade mode.
The project has allowed me to attempt to solve problems related to building an easily scalable word bank and level design with least manual effort. I used CEFR-tiered word selection (A1 to B2) to ensure the game respects the player’s vocabulary, creating a curated progression that gets tougher as you go.
On the technical side, it features an offline-first lexicon and a system designed for remote balancing. I can tweak the difficulty or the game economy without having to push a full update. I can now iterate on these details directly with the AI, figuring out the schema as we go.
🃏 Dice Poker: From Physical to Digital
The next one is a revamp of an older exploration I shared here in the newsletter. If you remember my post sharing a FREE Card & Dice game, you’ll recognize the core mechanic.
The original was a physical game with a standard deck, and a dice. Something about it worked in that setting. The tension of deciding which cards to hold, which to toss, was real. I wanted to find out if that could survive the jump to digital.
So I built it. The core loop stayed intact: 6 decks, 15 rounds, 5 rolls per hand. You pick which cards based on your luck with the dice and hope to build the strongest poker hand you can in every round.
The trickier part was the layout. Playing cards physically is different from playing on a phone where you’re usually holding it with one hand, scrolling with your thumb, half-distracted. I worked with Stitch to design the interface, sketching layouts in Figma first to give it enough context to work from. Getting from “this works in the physical world” to “this actually feels right one-handed” took more back-and-forth than I expected, but that process was the most interesting part.
🏏 Book Cricket Champion: The Nostalgia Trip
Finally, there’s Book Cricket Champion. This was my very first exploration with hosting on itch.io, and it’s actually an updated version of the game that started this whole AI experimentation journey.
For those who didn’t grow up in South Asia in the 90s, “Book Cricket” was the ultimate bored-in-class activity. You’d take a textbook, flip to a random page, and the last digit of the page number would be your score. 0 was a wicket, 8 was a single, and 2, 4, and 6 were the big hits.
I wanted to start playing with building something using AI, and this seemed like an easy design to try. In this version, you can play using content from over 100 literary classics, where every book uses its actual page count to determine the score.
This project was a pure “vibe coding” experiment. I used Google AI Studio to figure out the basic loop and then brought it into Antigravity to polish it. It features a local PvP mode so you can pass your device to a friend, exactly like we used to pass textbooks around. It’s a tribute to a simpler time, built with the most advanced tools we have today.
📈 Why Am I Doing This?
There’s value in building smaller prototypes even if they aren’t “production ready.”
In my opinion, AI is going to be an important part of our process, and as designers, it’s worth understanding the basics of how these tools work. From what I’ve tried so far, the speed jump is real — you can get from idea to something playable faster than I expected, though “easy” certainly doesn’t mean “automatic”.
I am perfectly aware that these games aren’t 100% production-ready. They aren’t going to top the App Store charts tomorrow, and I don’t intend for them to. That’s a task for the pros, the engineers and artists who spend years mastering their craft.
For me, these AI-assisted prototypes are the next step beyond pen and paper, or Miro and Figma presentations.
When explaining a design to a team, a static document can only go so far. A playable thing, even one with messy code under the hood, could communicate intent better than a long spec document. It lets you test the “feel” of a feature before a single line of production code is written.
I’m using these smaller experiments to build my own understanding of technology and, hopefully, get better at my skills as a designer. I want to continue trying more stuff and posting more prototypes here.
💭 The Feedback Loop
At the end of the day, the AI is just a tool for exploration.
When you play these, don’t look at them as finished products. Look at them as a window into a designer’s thinking process. I’d love to hear what you feel about them. Is the logic in Dice Poker too punishing? Does YAWG need more specific categories? Is the AI in Book Cricket too good (or too bad)?
Drop a comment on the itch.io pages or here. I want to know your thoughts. AI has compressed the development time, which means we can spend more time on the thing that actually matters: making the games fun.
Stay tuned for more. I’m genuinely curious to see where this goes next — and I’d love for you to come along for it.
📝TL;DR
Playable Things : A new section available on the newsletter featuring prototypes and design experiments.
AI as a Tool : I’m playing with AI tools to turn design intent into something playable.
YAWG : A word game focused on scalable level design and automated word banks.
Dice Poker : A digital version of a physical card game, built for mobile-friendly play.
Book Cricket Champion : A simple nostalgia-themed experiment to test AI-assisted building.
Design Communication : Using interactive prototypes as a possible step beyond static design documents.
It’s been a while since I posted. I’m going to blame it on these experiments, which is only partly true (the rest is just good old-fashioned procrastination). If you’ve played any of these, or you’ve been doing your own experiments with AI, drop a comment or reply to this. I’d genuinely love to hear what you’re building. And if you haven’t subscribed yet, what are you waiting for?
Thanks for reading Game Design Bites! Subscribe for free to receive new posts and support my work.
Top comments (0)