Low Code RAG-LLM Framework for Context-Aware Querying in Electrical Standards, Design, and Research

Large Language Models (LLMs) have demonstrated remarkable capabilities in natural language understanding and generation. However, in domain-specific applications such as electrical engineering, standalone LLMs are often susceptible to hallucinations and inaccuracies, limiting their reliability for technical querying. Retrieval-Augmented Generation (RAG) frameworks address this limitation by grounding LLM outputs in external, verified sources, thereby improving factual accuracy. Despite these advantages, most RAG implementations demand significant programming knowledge and infrastructure setup, creating a high barrier to entry for practicing engineers and engineering researchers who may not have a software development background. To address this challenge, this manuscript introduces a low-code RAG-LLM framework built using the N8N automation platform, enabling users to construct and deploy advanced RAG pipelines with minimal technical overhead. The framework supports enhancements such as reranking, contextual retrieval, and knowledge graph-augmented generation, all of which are implemented in a modular and reproducible manner. Extensive testing was conducted to evaluate the impact of each enhancement on retrieval accuracy and generation quality, using documents from National Fire Protection Agency (NFPA), IEEE Xplore Digital Library, and CIGRE. The strengths and limitations of these approaches are discussed in detail to guide practical adoption. All associated N8N JSON workflow files are made available with this manuscript to facilitate replication and customization by engineering professionals and researchers.