This workspace is where you review what agents are doing, what people are correcting, and which experiments should become the next operating baseline.