The integration of multimodal generative AI techniques in agriculture holds transformative potential, particularly in the automated detection and management of plant diseases. In this study, we present a novel approach that combines convolutional neural networks (CNNs) and contrastive language-image pretraining (CLIP) to develop a multimodal generative AI system for agricultural diagnostics. Utilizing the PlantDoc dataset, encompassing various crop diseases, we train multiple CNN architectures for disease detection and identify ResNet as the most suitable model for our dataset. The system predicts plant diseases, symptoms, and treatment recommendations, while the integration of CLIP enhances its diagnostic capabilities by enabling natural language queries and multimodal reasoning. This framework represents a scalable and efficient tool for mitigating crop losses and improving disease control strategies.