We present KEEL (Knowledge Engineering for Educators Layer), a multi-agent platform that empowers educators to organize and interact with fragmented instructional content. KEEL combines the semantic reasoning capabilities of Large Language Models (LLMs) with the dynamic retrieval capacity of Retrieval-Augmented Generation (RAG) to facilitate the extraction, structuring, and exploration of educational materials. Recent advances in large language models and retrieval-augmented generation have opened new avenues for assisting educators [3], yet effective automation tools for organizing course materials remain limited. The system supports the ingestion of heterogeneous document formats (PDF, DOCX, PPTX, TXT) and employs prompt-driven agents to segment content into coherent modular structures, such as courses, modules, and topics. Extracted knowledge is represented as automated Knowledge Maps, which serve as both a visual schema and a structured foundation for semantic retrieval. By indexing content in a vector database and enabling natural language interaction, KEEL supports context-aware question answering directly grounded in instructional artifacts. Existing systems that leverage LLMs or knowledge graphs for educational content tend to focus on full-text analysis or content generation. For example, domain-specific models like SciBERT and MatSciBERT excel at scientific text mining and relation extraction, and tools such as WorkedGen generate new worked examples via LLMs. Likewise, platforms like MagicSchool automate lesson planning and grading with AI. However, these approaches either require extensive content processing or produce isolated outputs, and none directly tackle the structuring of an instructor’s own teaching materials.