Video | LunarX Founder Yang Zhilin 2025 Interview: "The Management Team is Like Training a Large Model"

In August 2025, Yang Zhilin, the founder of the Dark Side of the Moon, shared a unique team management method during his participation in the "Zhang Xiaojun Business Interview." To manage an innovative team, one can draw inspiration from RL and SFT in AI training. SFT-style management involves directly telling employees what to do, while RL-style management involves setting goals and rewards, allowing employees to explore the implementation path on their own. The former has stronger constraints, while the latter offers more freedom.

According to Yang Zhilin, relying solely on SFT can easily suppress creativity, and relying entirely on RL may lead to the problem of "gaming the metrics." Therefore, the more reasonable approach is to primarily use RL with SFT as a supplement, finding a balance between free exploration and necessary constraints.