A Digital Twin of the road requires more than just 3D geometry. It needs a reliable virtual reference structure in which objects, sensors, states, and processes can be clearly referenced and linked to one another. Geometric-semantic models provide the foundation for this. They describe the physical road environment in an use-case-specific and semantically enriched way, with precisely considers the level of as-is detail required for analysis, assessment, and prediction within the defined system boundaries. This presentation introduces a process for generating such geometric-semantic as-is models in the context of the Collaborative Research Centre/Transregio (SFB/TRR) 339 “Digital Twin Road”. It shows how point clouds, existing data sources, and model-based processing steps can be used to create a virtual representation of the road that serves as a reference model for Digital Twins and Digital Shadows. The practical implementation is then demonstrated using a demonstrator for the use case “Monitoring of the Tire-Pavement Interaction”. Live vehicle positions, weigh-in-motion (WIM) data, and real-time AI-based deformation estimates are integrated in a game engine on a geometric-semantic road model. This creates a linked Digital Shadow that makes dynamic loading processes visible and provides a basis for future analyses, simulations, and decision-making processes in road infrastructure operation.

Building Information Modeling (BIM) authoring in the Architecture, Engineering, and Construction (AEC) sector entails navigating complex professional software, demanding significant manual effort and domain expertise. While current automation approaches rely heavily on rigid Application Programming Interfaces (APIs), they often detach agents from human-centric visual workflows, with API-specific implementations further limiting their generalizability across software platforms. In this work, we introduce BIMgent, a multimodal LLM-based agentic framework designed to perform autonomous BIM authoring directly via Graphical User Interface (GUI) interactions. To mitigate the challenges of high visual noise and long-horizon dependencies inherent in BIM authoring software, BIMgent incorporates a hierarchical planning mechanism guided by software documentation and employs dynamic GUI grounding to precisely navigate complex user interfaces. We evaluate our framework on a benchmark derived from real-world residential building designs. Experimental results demonstrate that BIMgent achieves a 77.8\% end-to-end success rate in correctly generating semantically rich BIM models from input floorplans, significantly outperforming the leading generalist computer-use models, including Claude-Sonnet-4.5 (2.2\%) and GPT-5.4 (11.1\%). By bridging the gap between abstract design inputs and concrete modeling actions without API constraints, BIMgent presents a viable and generalizable path toward autonomous architectural modeling assistants.