Chess Coach

I was inspired by the idea that chess practice should feel more like coaching than punishment. Most puzzle apps tell you whether you were right or wrong, but they do not explain the thinking behind the move in a way that feels personal. I wanted to build something that helps a player understand why a move matters, not just whether the engine approves of it.

I learned a lot about shaping AI into a useful teaching tool instead of a noisy feature. That meant controlling when hints appear, limiting repeated requests, and making the feedback feel responsive without overwhelming the user. I also learned how much polish matters in a learning product: timing, tone, and the order of information can change the whole experience.

I built the project as a chess training web app with puzzle flow, move review, and AI-generated hints. The app uses a modern frontend stack and connects to external AI providers for adaptive feedback. I focused on making the hint system only react after a move is made, so the AI acts like a coach reviewing your decision instead of a tool that interrupts the game. In the background, I added fallback handling and key rotation so the experience stays usable even when one provider is rate-limited.

The biggest challenges were rate limits, noisy fallback behavior, and avoiding extra AI calls when the user had not made a move yet. I had to rethink the request flow several times so the app would stay stable, avoid duplicate prompts, and keep the feedback relevant. Another challenge was balancing clarity and simplicity: I wanted the interface to feel helpful, not cluttered, so I removed extra panels and kept the hint area focused on one thing at a time.

A useful mental model for the project was that the app should maximize learning value while minimizing distraction, almost like optimizing ( \text{helpfulness} - \text{noise} ). That idea guided the final design of the hint experience.

Built With

• analysis
• api
• chess.js
• gemini
• html
• openrouter
• react
• stockfish
• typescript
• vite