I started Baron Buff because I wanted to build something interesting with AI and machine learning, and League felt like a good place to experiment. There is a lot of structure in the game, but also a lot of chaos. Drafts, matchups, team comps, player habits, timing, and game state all matter, and the “right” answer is usually contextual rather than obvious. It ended up being much harder than I expected. Solo building it from the ground up has been a real learning experience, especially once the project moved beyond “cool AI idea” into an actual product with a backend, desktop connector, real-time state, data pipelines, billing, deployment, and users in different countries. One of the biggest lessons has been learning where AI coding helps and where it hits limits. It can move fast on isolated tasks, but distributed systems are less forgiving. Race conditions, stale state, WebSocket timing, background workers, and production edge cases still require careful reasoning and a lot of debugging. The business side has also been more complex than I expected. Selling SaaS internationally means thinking about payments, taxes, localization, support, trust, pricing, and product messaging across different markets. None of that was obvious when I started. I’m still early, and there is plenty to fix, but that is also what makes it interesting. Baron Buff started as a way to explore AI and ML in a messy real-time game, and it became a crash course in building a real product end to end.

Baron Buff

League of Legends AI coach, companion app & LoL tracker

League of Legends AI coach, companion app & LoL tracker