Yeah, it was basically this, and running a whole lot of agents concurrently to churn through the dataset faster.