I recently hosted a Pokemon TCG prerelease event for my friends, most of whom were new to the card game. In typical developer fashion, I over-engineered the solution by building a small website instead of just using a phone timer and printed rules.
As the Pokemon enthusiast in my group, I wanted to introduce everyone to the Brilliant Stars set and make it as easy as possible for newcomers to learn. To create a fun and engaging experience, I built a mini-site featuring a retro-styled landing page, a tournament timer, and a handy rules reference—all developed with the help of AI coding tools.
Using AI as my development partner
I tested several AI coding tools for this project. My process was pretty straightforward:
- Wrote a detailed prompt describing what I wanted
- Refined it using ChatGPT to clarify requirements
- Used Copilot Edit to generate the initial code
- Iterated multiple times to match my vision
What I ended up with
The site had three main components:
1. A retro-styled home page
Simple landing page with event info and an ASCII art Pikachu that captured the nostalgic vibe I wanted.
2. A tournament timer
A dual-player timer that:
- Tracked 30-minute match time
- Kept track of each player’s 2-minute turn
- Entered “extra turns” mode when time ran out
- Had a clever UI with the top player’s controls rotated 180° for face-to-face play
It worked great on mobile, though we used it less than expected during the event.
3. A rules reference
This became the most valuable part - a clean reference with all essentials:
- Game setup
- Turn structure
- Win conditions
- Status conditions
- Tournament procedures
The AI even added emoji icons to make scanning easier. This was most used by my friends to be honest.
How the event actually went
The Brilliant Stars Set was a hit, with several friends pulling illustration rares from just eight packs each. The Lucario starter deck dominated our tournament, and I learned that the pre-constructed starters influenced strategy more than the booster packs.
The website proved most valuable as a reference document. I displayed it on my laptop connected to a TV while explaining deck building, and everyone referenced it during matches.
By the end, even newcomers were discussing type advantages and energy acceleration strategies - definitely the most rewarding part.
My takeaways about LLM-assisted development
A few key takeaways:
-
Natural language programming works – describing what I wanted and getting working code was genuinely useful. I was surprised at how well it worked.
-
Initial prompts matter – they significantly influence the overall design, making major changes difficult later.
-
AI handles boilerplate beautifully – CSS layouts, responsive design, and basic interactivity were no problem.
-
Visual design needs specific guidance – AI tools aren’t quite there yet for polished design without detailed prompts.
-
Human oversight is essential – I still needed to review content for accuracy.
I’ve shared this with friends who run Pokemon leagues, and they’re interested in using it for their events. Might add features like a pack simulator and tournament brackets in the future.
If you’re interested, check out the mini-site here (feel free to use it for your own events!)