This paper demonstrates possibilities and challenges in using an LLM to create a domain knowledge graph (DKG) for the work system domain and then converting it to a situation-specific knowledge graph (SSKG) for a specific system. It starts by summarizing existing ideas about work systems and a taxonomy of knowledge objects that have been presented previously. It uses an LLM to create a DKG based on parts of the work system perspective and highlights semantic challenges in that process. It uses an LLM to apply that DKG in creating SSKGs for two case study examples, one about ride hailing and one about medical care. It uses the LLM to identify parts of the DKG that were not the involved in links to nodes containing information from each case. It uses an LLM to prune those unnecessary nodes from the SSKG but also asks the LLM to identify parts of the DKG that might be relevant even though they were not included in a pruned SSKG. The conclusion stresses both practical possibilities and semantic challenges revealed in this research and discusses next steps related to use of LLMs in conjunction with DKGs and SSKGs in systems engineering.