A generative AI model for automating construction bid preparation

Construction bid preparation remains predominantly manual, labour-intensive, and time-pressured, often constrained by fragmented information, inconsistent documentation, and limited knowledge transfer. Despite advances in digital technologies, the construction industry still lacks effective AI tools to automate core bid preparation activities. This study develops a proof-of-concept Generative AI (GenAI) framework that automates key components of bid proposal development. A two-phase research design is employed: (1) a qualitative case study of two major engineering consulting firms in South Africa to understand existing workflows, challenges, and automation opportunities; and (2) the development of a GenAI system informed by these findings. Data were collected through semi-structured interviews with bid personnel and analysis of historical proposals and Requests for Proposals. The study identifies high-value automation targets, including technical write-ups, and translates these insights into a GenAI framework integrating LLMs, vector-based semantic retrieval, and structured bid repositories. Expert evaluation and a quantitative performance test (average F1 score: 96.25%) demonstrates the model’s ability to significantly reduce manual effort, improve consistency, mitigate knowledge loss, and enhance turnaround times. This research contributes a novel AI-driven framework for construction bid preparation and advances understanding of how GenAI can support operational efficiency, organisational learning, and sustainable knowledge management in construction procurement.