When I set out to build Proper Drop, the goal was simple: make an app to rate Guinness, and do it fast - there was a pub crawl coming up. I didn't want to get bogged down in configuring servers or wrestling with outdated frameworks. I wanted to rate pints, not manage infrastructure.
Here's a look under the hood at the chosen stack and the AI-assisted workflow that helped ship it quicly.
For the frontend, the old-school "Create React App" was diteched in favor of Vite. Vite uses modern browser features to serve code on demand, which meant we could tweak the rating slider or adjust the "beer foam" physics and see the changes immediately. That speed keeps you in the flow.
I didn't want to write a custom backend API. I chose Google's Firebase because it gives us authentication, a real-time database (Firestore), and hosting all in one box. The best part is the "Activity Feed." Because Firestore is real-time, when someone rates a pint at a bar downtown, it pops up on everyone else's feed instantly without us writing a single line of WebSocket code.
The beauty of this build wasn't just the stack, it was how everything came together. I am a Computer Engineer by training, but I still leaned heavily on AI tools like Perplexity to act as a force multiplier. Instead of Googling for hours to find the perfect SVG path for a beer glass, we could ask models like Anthropic's Claude or Google's Gemini to "generate a React component that looks like a pint glass and fills up based on a slider value."
This "Vibe Coding" approach allowed me to prototype complex features—like using the phone's accelerometer to make the digital beer tilt. This was done in minutes rather than days. It turns development into a conversation: we describe the feature, the AI scaffolds the logic, and we refine the UI to match our gold-and-black aesthetic.
It's a fun way to build software, and it let me ship a polished, production-ready app faster than I could on my own.