Right now I'm breaking down SpeedRun into two distinct phases.

Phase one? That was the original crew. Tight-knit group, got in before anyone else, rough around every corner. We weren't chasing airdrops—we were pressure-testing the whole loop: how prompts fed into data, how deployment actually worked, how rewards flowed back. Back then, Covalent was laser-focused on nailing down the fundamentals of that cycle. It was messy, but that's where the real learning happened.