Personalised learning systems are becoming essential in modern education due to the increasing demand for adaptive and context-aware knowledge delivery. However, most traditional learning platforms lack subject-level isolation and fail to effec-tively utilize user-provided study materials. To address these limitations, this paper proposes SAGE.AI (Subject-Aware Gen-erative Educational Agent), an intelligent system that integrates Retrieval-Augmented Generation (RAG) with Large Language Models (LLaMA), guided through structured prompt engineer-ing, to deliver subject- specific and context-aware responses. The system allows users to dynamically create subjects, upload PDF-based learning resources, and interact through a dual-interface design consisting of a PDF analysis panel and a chat interface. The RAG pipeline extracts and processes relevant content from uploaded documents, while carefully designed prompts ensure that the LLM generates responses strictly constrained to the selected subject. This approach enforces subject isolation and prevents cross-domain response leakage. By combining document understanding, retrieval-based context injection, and controlled generative responses, the proposed system enhances learning efficiency and personalisation. SAGE.AI can be deployed as a scalable web-based educational assistant for students, profession-als, and self-learners.
R/ME et al. (Thu,) studied this question.