Precision agriculture is an important component of increasing crop productivity through data-enabled decision-making. We present a hybrid approach that employs deep learning (DL) and ensemble learning to enhance crop disease classification and yield forecasting. Our model employs a hybrid convolutional neural network (CNN) architecture that combines the EfficientNetB0 and InceptionV3 models, which leverage their high-performing feature extraction performance to classify common corn, rice, and wheat diseases. Once a crop has been determined to be healthy, the model forecasts yield using a random forest model based on the defined plot of land. The proposed two-stage process provides certainty in disease classification and yield estimation in support of the decision-making process of farmers. The findings indicate that the hybrid model achieves an accuracy of 98.4% for disease classification with a 15% increase in yield estimation accuracy when compared to baseline methods. In general, it was noticed that selecting deep and ensemble learning resulted in reliable and scalable AI-based solutions to precision agriculture. Overall, the research supports sustainable agriculture with an intelligent decision support system for both disease mitigation and yield forecasting.