Conrad, known in the League of Legends scene as stunner, joins Sunny Ray to discuss GamingAcademy.gg (GA.gg), the German esports e-learning platform he built to make coaching affordable and scalable using AI. Conrad explains how a personal frustration with mediocre coaching led him to launch a side project in 2017 that became a company in 2021, bootstrapped largely through business angel funding rather than venture capital. He details HexCore AI and its mascot Kochi, which analyze Riot Games API data to identify player weaknesses and recommend personalized training videos. Conrad is candid about where AI content generation fails, describing failed experiments with AI dubbing and lip syncing that broke down on fast video cuts. He argues that durable products blend AI efficiency with human quality control rather than chasing full automation. The conversation also covers scaling limits at the highest skill tiers, where human coaches still outperform AI, and how Conrad is now expanding into a B2B broadcast arm for Riot Games while pursuing global growth for the platform.
An esports founder explains why AI coaching scales beautifully until you hit the ceiling only humans can reach.
What are you building right now?
We built GamingAcademy.gg, GA.gg for short, named after the esports phrase good game. We replace the cost and time of human coaching by analyzing petabytes of data from the Riot Games API for each user individually, identifying weaknesses and telling players exactly how to improve and win more games in League of Legends, the only title we currently cover.
How did competing in League of Legends turn into a business idea?
I was a mediocre player searching for coaches and e-learning platforms, but everything I tried was bad. I realized I could build something better myself, especially with new AI opportunities, offering a cheap monthly price for guides, statistics, a specially trained chatbot, and a community, everything a player needs in one place.
Why did you bootstrap instead of raising big venture capital?
It started like the playbook says, pitching to a friend who knew angel investors. We got about 70k euro externally plus our own money, starting with 100k total. We built an MVP without AI first, just video coaching content, to prove people would pay before adding AI and statistics later. In total we raised about 970k euro from business angels.
What convinced you esports coaching could be productized and scaled?
Many gamers are stereotypically people who do not want to interact with a human coach for an hour, they want videos and statistics on demand. Live coaching also costs 25 to 30 euro per hour and requires scheduling, sometimes days out. Removing that inconvenience and cost, letting people learn immediately whenever they want, is what convinced me.
What does Kochi and HexCore AI actually do compared to a human coach?
It does exactly what a human coach does, like a tennis coach spotting your bad backhand, except we do it through data instead of sideline observation. We compare a player's stats to others at the same rank, use AI to find which skill gap would most improve their win rate, then recommend the right video to fix it specifically.
Where does AI coaching genuinely fall short compared to human coaches?
Above a certain skill level, roughly master rank or higher in League of Legends, which is under the top 1 percent, a human coach becomes more valuable. Our defined skill framework does not cover problems outside it, and we are designed for the mass market below that tier, not for players chasing a professional career.
What broke the illusion that AI could fully replace human video production?
We tried AI dubbing tools to translate premium courses, generating English voice and lip sync from German footage. It worked fine for a static talking head but broke on our style of fast cuts between the speaker and in-game footage, sometimes animating a champion's head to speak instead of the presenter. We had to record separate German and English versions instead.
What's the one hard lesson you'd teach another founder about AI content?
Do not try to generate your content purely with AI. Use it for specific narrow cases, like removing part of a background from a real picture, but always start from something real rather than generating a scene from nothing. Users can feel when content is genuinely made versus AI slop, and gaming communities are especially unforgiving of that.
Building something daring? Sunny talks to founders like this every day. Fifteen minutes to see if your story belongs on the stage.
Claim your pre-interview