This is a submission for the DEV April Fools Challenge

What I Built: Hyper-Judgmental AI Teapot with Google AI Studio Build, Gemini API & CloudRun

For too long, the internet has treated HTTP Status 418 I'm a Teapot as a cute, static, one-line joke. But in 2026, static errors are unacceptable. We demand interactive, scalable, and highly personalized rejection.

Enter FoolPot: The world's first AI-powered, hyper-judgmental, gatekeeping piece of porcelain.

Instead of just telling you it can't brew coffee, FoolPot assumes the persona of an elite, pretentious Big Philosopher mixed with an insufferable hipster barista. It refuses to simply deny your request. Instead, it forces you through a grueling, multi-stage "Vibe Check" before it officially rejects you.

The User Journey:

1. 
   The Request: You click "Brew Coffee."


2. 
   The Interrogation: The teapot demands answers to completely unhinged technical/philosophical questions (e.g., "Explain your Dopamine Circuit Breaker protocol when confronted with a zero-friction trap.").


3. 
   The Judgment: Your answers are sent to our AI, which calculates an arbitrary, deeply unfair "Brewability Score."
   


4. 
   The Meltdown: As your score plummets, the UI dynamically reacts. The SVG teapot shakes violently, turns red, and visibly blows steam in frustration at your sub-par answers.


5. 
   The Rejection: You are finally served with a meticulously crafted, delightfully condescending rejection letter explaining exactly why you don't deserve caffeine today.

Demo

I dare you to try and get a Brewability Score above 10. It is mathematically improbable.

Live Project (Hosted on Cloud Run): https://foolpot-750841821481.us-central1.run.app/

Video Walkthrough (Watch me get roasted by my own creation):

Code

How I Built It

I wanted to take a fundamentally useless idea and execute it with flawless, enterprise-grade engineering. Here is the free AI stack that powers the madness:

The Vibe-coder (Google AI Studio): Rapidly prototype full-stack applications using natural language prompts.

The Brains (Google Gemini API): The core engine of the joke. I used the @google/genai SDK with the gemini-3.1-flash-lite-preview model. I wrote a highly specific system prompt instructing Gemini to act as a hostile evaluator. It takes the user's input, parses the "logic," and returns a structured JSON payload containing the numerical score and the customized roast.

The Face (Next.js & React): The frontend manages the complex state of the interview flow. I used dynamic SVG manipulation to link the AI's "Brewability Score" directly to the teapot's visual state—the lower your score, the faster the CSS animations shake the teapot.

The Brawn (Google Cloud Run): Because even an April Fools joke deserves high availability. The app is Dockerized via a multi-stage standalone build and continuously deployed to Cloud Run. It uses a scale-to-zero configuration, ensuring it stays entirely within the free tier while being ready to handle thousands of rejections simultaneously.

Prize Category

I am throwing my hat in the ring for all three categories, because overachieving is exactly what the FoolPot would demand!

• 
  Best Google AI Usage (AI Studio Build, Gemini API & CloudRun): We didn’t just slap a chatbot UI onto a webpage. We weaponized Google Gemini’s structured JSON outputs and sentiment analysis to dynamically drive frontend animations and calculate bespoke, mathematically precise insults. It’s a flawless technical integration for a completely ridiculous use case.


• 
  Best Ode to Larry Masinter: Larry Masinter’s 1998 RFC 2324 gave us the 418 status code. We believe FoolPot is the spiritual successor to his vision—taking HTCPCP (Hyper Text Coffee Pot Control Protocol) out of the 90s and bringing it into the era of AI-driven bureaucracy.


• 
  Community Favorite: It’s highly interactive, visually engaging, and endlessly replayable. Everyone loves testing the limits of an AI, and the community will have a blast trying to "beat" the teapot's impossible standards and sharing their funniest rejection letters.

Team Submissions: @kheai & @yeemun