An innovative mango leaf disease detection system was designed using advanced deep learning technology integrated with interactive user support. A state-of-the-art multi-optimization CNN model was developed to accurately classify mango leaves into eight possible states: anthracnose, bacterial canker, weevil, dieback, gall, powdery mildew, sooty blight, and healthy. Utilizing four optimizers – Adagrad, SGD, Adam and RMSprop – the model achieves a remarkable accuracy ranging from 98.53% to 99.47% in various test scenarios. This approach leverages the strengths of each optimizer to improve the robustness, adaptability, and reliability of disease detection and efficiently solves problems presented by the different characteristics of datasets. To further improve usability, an artificial intelligence chatbot interface has been implemented that uses state-of-the-art tools developed by OpenAI. This feature enriches the system by offering real-time disease diagnosis along with tailored preventive measures and treatment recommendations for agricultural solutions. The platform's responsive interface allows users to continuously contribute by uploading new data, allowing the system to refine its capabilities and improve its performance over time through continuous learning. This innovative solution outperforms existing disease detection methods and offers significant potential for large-scale applications in precision agriculture and automated disease management systems. By making machine learning accessible and intuitive to end users, this system represents a significant leap forward in agricultural technology. It facilitates early detection and treatment of mango leaf diseases and paves the way for smarter, data-driven and sustainable agricultural practices.