Adam Liu & Jai Das & Adrian Kania — Sapphire Ventures - Reinforcement Learning: Learning by Doing - April 2026

An article from Sapphire Ventures explaining reinforcement learning (RL) as a training method for AI agents, emphasizing learning by doing. It details the four key components of RL systems: environment, task, verifier (grader), and feedback loop, using an analogy of learning golf. The piece covers how RL differs from supervised learning and highlights its importance for building capable AI agents.