Knowledge Graphs (KG) have risen to be a powerful mechanism to represent data. Despite this most data sources are generally still represented via heterogeneous non-graph data structures. Converting these into KGs necessitates considerable effort from experts, proving this to be a time consuming process. While tools have been developed to aid KG builders, a gap still exists in terms of technologies that support the automation of designing KG building pipelines. Addressing this gap motivates this research. The aim is to first understand the problem at the knowledge level and, inspired by the recent release of generative tools such as GPT-Engineer, to put forward a conversational agent aimed at assisting the user in building their pipelines. We report on the preliminary findings that we have so far reached during the first year of research in deriving the requirements for building KG generating pipelines from the literature.