Machine-learning approaches are providing new insight into bird comfort during loading and transportation, which could reduce losses due to overheating.