General Reasoning Co-Founder and CEO Ross Taylor explains how they are using Reinforcement Learning on teams of agents rather than 'individuals': "We've just launched our new model series, Avalon M1. The 30,000-foot pitch is: a lot of agents in the past were kind of trained as like single models, but we've essentially developed a recipe to do reinforcement learning on groups." "Instead of training like a single smart agent, you train like a team, an organization, eventually like an economy or…