Chessray is an AI-powered chess coach that analyzes a player's last n chess games and generates personalized coaching advice to help them improve at chess. It was inspired by my time training with a Chess Grandmaster to improve my own game. Launching soon.

Most chess analysis tools focus on single games - they tell you what went wrong in a single game. But club players like myself (rated 1400-2200) plateau because we keep making the same mistakes across dozens of games. ChessRay solves this by finding recurring patterns across your recent games and generating personalized coaching advice to break through.

Growing from 1,800 to 2,000 ELO

My Chess.com rating progression

I've been playing chess for the last 5 years. Lately though, I've felt that my growth in the game has been quite slow. I reached out to the Reddit community for tips on improving my Chess ELO (a global rating system for chess players) from 1,800 to 2,000+.

One of the main suggestions I received was to zoom out and do a thorough meta-analysis of my own chess games. To analyze not only one chess game at a time, but my last 20+ games. To understand the strengths/weaknesses that are inhibiting me from growing as a player.

The key to 2,000+ ELO, I learned, is to spot weak areas and intentionally design my daily practice around them. With any endeavour, deliberate practice is often better than meandering practice.

It was about asking the same types of questions that my GM coach often asked me:

• Where do I usually lose advantage?
• What phase of the game do I struggle with most?
• Do I repeat certain types of mistakes often?

But knowing what to work on isn't easy. We can't always see our own blindspots. This is why coaches can unlock growth; they help us see from a vantage point what we often miss in our self-critique. Doing a meta-analysis of my last 20 chess games would take me 3-5 hours at least, and be limited by my current knowledge. Doing this every 2 weeks for continued growth seemed like a daunting task, as a hobbyist chess player.

Building the tool

So, I spent a few hours over the weekend to build a tool for myself that does exactly this, in less than 1 minute:

• Pull my last 50 chess games
• Analyze all games through a chess engine
• Breakdown performance by opening, middle and endgame
• Find critical, game-defining positions
• Suggest opening lines to improve my win-rate
• Highlight strengths and weaknesses in my game style
• Build a personalized plan for me to reach the next 100 ELO

I suddenly had a mini chess GM coach on my terminal.

The Analysis

A v0.1 iteration

The first report was mind-blowing. The analysis was painfully accurate ("practice rook endgame fundamentals" 🥲), and it felt like magic. It unlocked patterns that I had been missing for months. I learned why I've been stuck, and what to spend time on. It felt having a WHOOP, but for chess.

Technical Architecture

I settled on a 4-stage pipeline to answer this question:

Early Feedback

I proceeded to launch it on Reddit, and overnight had 50+ people who wanted to use the tool too.

What's next

This is just a V1, but I'm already brimming with so many more ideas to grow it. This is now a part of my toolkit on my side-quest of becoming a titled chess player.

This has the potential to become a valuable tool for serious hobbyist chess players. Launching soon as a standalone product.