This study introduces an AI-driven approach for real-time query management and FAQ(Frequently Asked Question) generation, addressing the challenges of scalability and immediate responsiveness in large-scale online learning environments. During live sessions, student queries are embedded and evaluated against the FAQ knowledge base using a dissimilarity metric. Queries exceeding a predefined dissimilarity threshold are escalated to instructors, while lower-dissimilarity queries are addressed by a Retrieval-Augmented Generation (RAG), optimized to ensure minimal latency and high accuracy in real-time responses. We build the RAG by embedding course material into a high-dimensional vector space, organizing it through topic clustering, and constructing an FAQ database pre-validated by instructors. The system achieves a notable answer relevancy score of 88.65% and a contextual precision score of 88.25% underscoring its balance of precision and adaptability.