Microbe Desk

• 

  Game Bench

• 

  Game Icon

• 

  Gram-Stain Example

• 

  Main Menu

Inspiration

Microbe Desk began as an original iPhone game built in Xcode with Swift, created entirely with ChatGPT assistance. Its original vision was a tactile microbiology investigation game: players build a case from lab results, patient information, and physical documents spread across a laboratory bench.

I pivoted it into a web app because I wanted anyone to be able to play directly in a browser—without needing an iPhone, an installation, or prior microbiology knowledge—while preserving the original hands-on investigative feeling and making it more accessible as an educational experience.

How we built it

I used ChatGPT to build the original Swift/Xcode game, then used Codex with GPT-5.6 to translate and expand it into a modern web experience built with React, Next.js, TypeScript, and Vite/Vinext.

The game contains a microbiology catalog, patient cases, tests, treatments, progression, and scoring as deterministic web data. Its centerpiece is a persistent evidence bench built with browser-native pointer, touch, keyboard, focus, and local-storage interactions. Players can drag, stack, read, file, and revisit documents while solving cases.

Microbe Desk is an AI-native build. Its visual assets—including educational Gram-stain and culture reconstructions—were generated with ChatGPT image-generation tools. Its original adaptive music and interface sounds are synthesized in the browser with the Web Audio API, using an original system created for the game rather than imported recordings, samples, loops, or third-party music.

I also added an optional server-side AI mentor. It provides personalized learning feedback only after the game has calculated a local verdict; it cannot alter medical facts, scoring, or progression.

Accomplishments that we're proud of

• Successfully pivoting an original Swift/Xcode game into a browser-based educational experience.
• Building an interactive evidence-bench concept instead of reducing the game to a multiple-choice quiz.
• Creating the game’s visual identity and educational lab-image reconstructions with ChatGPT generation tools.
• Building original adaptive browser audio without imported music or samples.
• Making microbiology approachable for beginners while retaining a meaningful reasoning loop.
• Using AI transparently: deterministic game data controls the facts and grading, while GPT-powered feedback supports reflection after each case.

Built With

• accessibility
• chatgpt
• cloudflare
• codex
• education
• game-development
• gpt-5.6
• microbiology
• next.js
• openai
• openai-api
• react
• typescript
• vite
• web-audio-api