Abstract
Representing systems using the SAPPhIRE causality model is found useful insupporting design-by-analogy. However, creating a SAPPhIRE model of artificialor biological systems is an effort-intensive process that requires humanexperts to source technical knowledge from multiple technical documentsregarding how the system works. This research investigates how to leverageLarge Language Models (LLMs) in creating structured descriptions of systemsusing the SAPPhIRE model of causality. This paper, the second part of thetwo-part research, presents a new Retrieval-Augmented Generation (RAG) tool forgenerating information related to SAPPhIRE constructs of artificial systems andreports the results from a preliminary evaluation of the tool's success -focusing on the factual accuracy and reliability of outcomes.