Making Generative Artificial Intelligence (GenAI), such as Large Language Models (LLMs) or Diffusion Models, accessible to public administration is an important task, because of the often textual and image-based activities in this sector. Such activities include the generation of documents, summaries, speech text, or explanations of large collections of documents, such as legal documents about laws, their interpretations, and judgments. One approach for the targeted use of domain-specific documents, e.g., documents of a collection, is Retrieval Augmented Generation (RAG). Simply speaking, for a prompt given to an LLM as a query, RAG augments it with selected text taken from a previously prepared and provided data set. However, this preparation greatly depends on the application and task to be performed. Especially for legal text, a query could simply be ”Which law deals with building modernization”, which would result in less structuring, a query such as ”In which cases does a tenant have to pay modernization costs?” would result in the need of higher structuring legal text. In this paper, we take a detailed structuring approach for enabling answers about legal text related to building modernization questions which come up when, e.g., an external thermal insulation system shall be installed on a front side of a building. This research is part of the Intelligent Modernization Platform (IntelMOD) project, which aims to offer a tool to tenants and landlords to initiate modernization negotiations based on building functions: stability, moisture protection, thermal insulation, sound insulation, fire protection, daylight access, aesthetics, and cost effectiveness.