The Public Administration domain can greatly benefit from integrating Generative Artificial Intelligence into its workflow, making daily tasks more accessible and less time-consuming. Large Language Models can prove beneficial in supporting the procuring entities and citizens in decision-making and accessing information. However, knowledge encoded in Large Language Models is bound to training data, and these data can become obsolete. These data-wise limitations are crucial in a dynamic domain like Public Administration, where regulations and guidelines are subject to frequent changes. To overcome these issues, Knowledge Graphs provide structured, directly editable and accessible knowledge. Nevertheless, Knowledge Graphs often suffer from incompleteness and incorrectness, and require extra curation. Large Language Models can help address these issues by expanding Knowledge Graphs and extracting new facts from unstructured documents. Preliminary experiments in our previous work OIE4PA, an Open Information Extraction framework to identify facts from Italian Public Administration tenders, showed promising results. Given this encouraging starting point, the objective of this research is to follow in OIE4PA footsteps, studying and developing a methodology to achieve a synergy between Large Language Models and Knowledge Graphs for Question Answering and Query-Focused Summarisation tasks, and whether or not it benefits the Public Administration domain.