V-Model Approach in Generative AI System Architecture Design for Construction Cost Management

The UK construction sector has increasingly encountered cost management inefficiencies in overruns, errors, rework and variations. This study demonstrated how Generative AI (GAI), an emerging trend in digital construction, can foster large language models (LLMs) from the industry's historical data to predict costs. The process of developing the GAI system architecture applied the V-model and agile methodological approach and BIM templates. As used in the UK construction sector, the BIM templates considered data from building cost information service (BCIS) and task information delivery plan (TIDP) to develop the architecture. The system architecture designed in this study aligned with the RICS New Rules of Measurement 1 (NRM1) for early cost advice and text-to-task models. The implication of the GAI system architecture for digital cost management presented in this study elicited the integration of GAI with the BIM processes, offering substantial benefits to the construction industry. This includes streamlined workflows, reduced errors, and improved decision-making. The implications of the system architecture offer opportunities for increased BIM uptake in the UK and the sector globally.