Who’s That Pokémon 🟡⚡

Inspiration

We wanted to build a game that blends nostalgia, knowledge, and modern web engineering. Pokémon is universally recognizable, and guessing-based games naturally drive curiosity and retention. The idea was to go beyond a simple quiz and create an experience that feels progressively challenging, replayable, and intelligent, while also serving as a strong showcase of frontend, backend, and AI-driven logic.

What it does

Who’s That Pokémon is a web-based guessing game where players identify Pokémon using progressively revealed clues. The game offers multiple modes such as silhouette guessing, Pokédex clue deduction, stat-based puzzles, and daily challenges.

Key features:

• Progressive hint system (sprite, type, stats, generation, clues)
• Daily challenge shared across all users
• Scoring based on speed, accuracy, and hints used
• Streaks and leaderboards
• Intelligent fuzzy guess handling
• Multiple game modes for casual and competitive players

How we built it

• Frontend: Next.js with Tailwind CSS for fast rendering and responsive UI
  
• Backend: Node.js with API routes handling game sessions and validation
  
• Data Source: PokéAPI for Pokémon stats, sprites, types, generations, and descriptions
  
• State Management: Client-side state for gameplay + server-side validation for fairness
  
• Caching: Aggressive caching of PokéAPI responses to reduce latency and API load
  
• AI Layer:
  
  • Smart hint generation
    
  • Fuzzy guess matching
    
  • Adaptive difficulty based on player performance
    

The game logic was designed as a pure engine layer, separate from UI, making it easy to add new modes without rewriting core systems.

Challenges we ran into

• Preventing name leakage in Pokédex descriptions while keeping clues meaningful
  
• Balancing difficulty so the game is neither trivial nor frustrating
  
• Handling fuzzy guesses without accepting incorrect Pokémon
  
• Ensuring fair play for daily challenges (server-authoritative Pokémon selection)
  
• Managing PokéAPI rate limits and inconsistent data across generations
  

Accomplishments that we’re proud of

• Built a scalable game engine that supports multiple modes
  
• Implemented adaptive difficulty without heavy ML infrastructure
  
• Designed a daily challenge system that encourages social sharing
  
• Achieved high replayability using simple mechanics and data-driven logic
  
• Created a Pokémon fan project that is IP-safe and API-first

What we learned

• Small gameplay tweaks (hint order, scoring curves) have huge impact on retention
  
• Caching and preprocessing API data is critical for performance
  
• AI doesn’t need to be complex to feel intelligent—good heuristics go a long way
  
• Separating game logic from UI drastically improves maintainability
  
• Constraint-based design leads to more creative solutions

Built With

• aistudio
• geminiapi
• nanobanana
• pokemonapi