With the fast growth of artificial intelligence (AI), numerous significant improvements have occurred in the healthcare industry, among them are automated diagnosis and patient interactions. However, symptom-based diagnoses have traditionally been carried out using approaches such as decision trees and support vector machines (SVMs), which are limited in flexibility and conversation-like ability essential for better engagement with patients. A new way of making better healthcare chats will be discussed in this study, where powerful large language models (LLMs) are used within the LangChain framework. Instead of using conventional machine learning models, we suggest using a language model-based approach to enhance diagnostic accuracy and improve user experience. Our methodology entails deploying a patient-reported symptoms chatbot system based on an LLM that generates possible diagnoses and any necessary information about health. We evaluate this novel approach against conventional ones on performance metrics such as diagnostic correctness, coherence of speech, and customer happiness. This study advances the growing field of artificial intelligence in health care through the demonstration of an effective implementation of state-of-the-art language models in systems that interface with patients. In subsequent research, we will aim at improving medical precision in this system, enriching its knowledge base, as well as assessing real clinical scenarios. The current research lays the foundation for advanced user-oriented AI wonderlands within this field, which promise better levels and accessibility to medical advice or assistance.