New Frontiers in Built Heritage Management. AI assistant for advanced data integration and building information retrieval

The recovery, valorisation and management of the built heritage require advanced systems capable of handling the uniqueness and complexity of heterogeneous data and multiple sources. While current approaches have improved the visualisation of geometric and non-geometric information, the reasoning and inference tasks still rely on structured knowledge and query through computer programming expertise. This paper explores the integration of an AI-driven OpenAI API Assistant with Heritage Building Information Model (HBIM) datasets and unstructured data sources, such as historical documents. The developed workflow begins with data collection and preparation, followed by AI Assistant development and, finally, a qualitative assessment. The proposed methodology is applied to the case study of the Sanctuary of Hercules and the former Segrè papermill in Tivoli, near Rome, to illustrate how this approach improves data accessibility, interpretation and interdisciplinary collaboration while addressing heritage management limitations as unstructured data organisation and resource-intensive processes. The results shows the improvements of a scalable and adaptable solution for querying complex data and information.