Abstract
One of the key challenges of Energy Management and Information Systems in buildings is related to the lack of interoperability, due to the absence of standardization of the underlying data models. In recent years, there has been a growing interest in using ontology-based metadata models to address this issue, as they offer a structured approach to organize and share information across diverse systems (e.g. Brick ontology). However, the creation of ontology-based metadata models is often a labor-intensive task that requires specific domain expertise, hindering the practical use of such data models. For this reason, in this work the BrickLLM Python library is introduced, which addresses this issue by generating Brick-compliant Resource Description Framework graphs through Large Language Models, automating the process of converting natural language building descriptions into machine-readable metadata. The library supports both cloud-based APIs (e.g., OpenAI, Anthropic, Fireworks AI), local models (e.g. LLaMa3.2, etc.) and evenfine-tuned ones. This paper explores the architecture, key functionalities, and practical applications of BrickLLM, showcasing its potential impact on the future of building systems monitoring and automation.