A new machine-learning model for predicting crop yield using environmental data and genetic information can be used to develop new, higher-performing crop varieties.