Adopting Large Language Models (LLM) has markedly improved the user assistance system. LLM-based chatbots such as ChatGPT, Gemini, andMetaAI are extensively utilised for customer support. This nascent technology is gaining prominence across many sectors, including education. Chatbot technology can deliver prompt and customised services to all individuals within the system. LLM-based chatbots are susceptible to hallucination due to occasional inaccuracies in the given content. Retrieval Augmented Generation (RAG) is an exceptionally effective solution to this issue, especially in knowledge-intensive fields such as education, where responses must be correct and precise. This article presents a chatbot designed using LLM and RAG to assist students at the university. Quantitative parameters like BERTScore and ROUGE score, as well as qualitative factors, including hallucination and accuracy, evaluate the chatbot’s performance.