Case Study
Tourist Trap Together
AI-Directed Game Creator · Work in Progress · Jun 2026 - Present
Overview
Created Tourist Trap Together, a work-in-progress browser multiplayer game where players can start a single-player route or join multiplayer sessions directly in the browser.
Key Technologies
Achievements
- Built the initial playable web experience over a weekend.
- Directed the full build with AI coding agents and did not manually write source code line by line.
- Shipped a public browser demo with a polished main menu, single-player route entry, multiplayer entry, settings, and account flow.
Story & Process
Tourist Trap Together is a work-in-progress browser game built around a playable online multiplayer direction. The project presents a stylized tourist-road-trip game world where players can start a single-player route or choose/create a multiplayer server directly from the browser.
The most important part of this project is the build process: I made the initial version over a weekend using AI coding agents. I did not manually write the source code line by line. My role was to define the product direction, guide the implementation, review behavior, refine the UX, and keep the system moving toward a playable browser experience.
What It Is
- A browser-playable game with a polished main menu and route-based game framing.
- Multiplayer-oriented gameplay direction, with account and co-op entry points.
- A Babylon.js-powered client intended for interactive 3D game experiences.
- A work-in-progress product that will receive more gameplay, multiplayer, and content updates over time.
Technical Direction
The project uses a modern browser game stack: React, Vite, TypeScript, Babylon.js, and react-babylonjs on the client side. The multiplayer/backend direction is supported by Colyseus, Express, Prisma, and PostgreSQL.
The architecture is designed around a web-first game experience, with scripts for web development, server development, full-stack local development, and desktop experimentation. The details are presented here at a high level only, without exposing implementation code.
AI-Directed Development
This project is part of my current experiment in building faster with AI coding agents. Instead of treating AI as autocomplete, I used it as an implementation team: I described the product, reviewed outputs, corrected direction, checked the running demo, and iterated until the experience felt presentable.
The result is not positioned as finished. It is an ongoing project and a proof of how quickly a weekend build can become a public, playable product when product direction, technical architecture, and AI-assisted implementation are combined well.