Agricultural diseases exert a profound impact on both crop yield and quality, leading to substantial losses in global food production. Consequently, the need for timely disease recognition has become increasingly evident throughout the cultivation process. However, a major challenge lies in the limited expertise of farmers to identify diseases at an early stage. Furthermore, even if changes in crops are detected, there may be a lack of knowledge regarding appropriate remedies. Resorting to expert consultations for on-site assessments can be time-consuming, risking the widespread transmission of infectious diseases. In order to bridge the knowledge gap between experts and farmers, the utilization of a conversational agent, which can provide continuous multilingual personalized support, has emerged as a promising avenue. Thus, this paper provides a comprehensive overview of existing agricultural chatbots, detailing their architectural designs and methods for handling the required knowledge. Recognizing a significant gap between the requirements for a personalized multilingual chatbot for plant disease detection and existing solutions, the study proposes a hybrid model that integrates an ontology-based knowledgebase, open-source frameworks, and a large language model-based architecture to develop a more intelligent chatbot for detecting paddy diseases in Sri Lanka. The ultimate goal of this work is to develop a voice-enabled agricultural assistant, promoting inclusivity and user-friendliness for farmers with limited digital literacy in the future.