This article presents a framework that embeds LLMs into Failure Mode and Effects Analysis (FMEA) to speed up and improve risk identification and ranking. The authors design structured prompts and collaborative workflows so that the LLM proposes failure modes, effects, and causes, which are then reviewed by engineers. They show that prompt templates and few‑shot examples significantly influence the quality and completeness of generated FMEA content. For PSM, FMEA is closely related to equipment and process risk analysis; the paper offers concrete patterns for prompt engineering in systematic risk studies, including how to structure inputs (system description, functions, known failures) and outputs (tables, rankings) in a way that can be adapted to process hazard and reliability analyses.
