Kakomi

I'm Damo. I play Go and have followed Lee Sedol's career for years. When he appeared on The Devil's Plan, I watched the show. He left early, but I stayed — and the Wall Go game caught my attention. I'd never seen it before.

I looked for somewhere to play it online. A few existed, but none had a strong AI opponent or enough players for a real match. So I built my own.

I wanted to see if the approach behind AlphaGo Zero — self-play and neural networks — could produce a strong Wall Go player. The AI trains against itself and improves over time.

The model is still training. The “expert” difficulty you can play against today is an early checkpoint — a test of the approach, not the finished product. Training takes compute I don't have much of, so progress is slow. But it plays a real game already, and it will keep getting stronger.