Zhang, L. and Ge, Q. (2026) Offsite construction risk management agent: A gpts based analysis model. Engineering, Construction and Architectural Management, 33(7), pp. 5625-5644. ISSN 0969-9988
Abstract
Purpose – The purpose of this paper is to introduce the Off-Site Construction (OSC) Agent, a prompt-based chatbot utilizing Generative Pre-trained Transformer (GPT) technology, design to manage risks in OSC projects. This tool reduces the need for extensive data and professional expertise, addressing key barriers in traditional risk management (RM) approaches. The research aims to address the unique and complex risks in OSC by providing a tool that helps participants in managing these risks without requiring extensive RM expertise. Design/methodology/approach – The OSC Agent was developed by integrating GPTs with a focus on three core abilities: interface design, prompt configuration and data integration. The OSC Agent development process involved analysing 25 in-depth interviews, and NVivo was employed to code and synthesize the qualitative data, ensuring a systematic identification of OSC-specific risk factors. The OSC Agent provides two primary functions: risk definition and risk analysis, an ablation study with 2, 379 queries validated the risk definition validation, these queries were derived from a structured RM dataset specifically created based on common OSC risk factors. Expert evaluation with 3 experts is used for risk analysis validation. Findings – The OSC Agent achieved a 99.95% accuracy rate in risk definition queries and an 88.6% accuracy rate in analysing risk factors from interview transcripts. The ablation study highlighted the importance of integrating the instruction and knowledge modules, enhancing accuracy and consistency. Expert evaluations confirmed the agent's ability to streamline the OSCRM process, reducing the need for extensive RM knowledge among OSC participants. Originality/value – This research presents the development of an innovative and versatile RM tool for OSC projects. The OSC Agent assists users in managing risks without requiring extensive RM experience, effectively reducing issues related to quality, cost and delivery in OSC projects.
| Item Type: | Article |
|---|---|
| Uncontrolled Keywords: | generative pre-trained transformer; large language models; off-site construction; risk management |
| Index terms: | risk management, evaluation, configuration, off-site construction, integration, risk analysis, large language model, validation, methodology, accuracy, dataset, interview, agent, module, risk factor |
| Subjects: | professional development, practitioner, building construction, environmental hazards, systems engineering, risk assessment, data collection methods, data management, data analysis and analytics, organizational analysis, data science, architectural elements, research methods |
| Topics: | Digital Applications, Research Practice, Information Management, Roles and Professions, Design Practice, Risk Management, Organizational Design, Sustainability, Construction Technology, Engineering Principles |
| Descriptive scope: | 5 PCTEA |
N.B. Descriptive scope is a count of how many of the five facets of empirical research are indicated by the words used in title, abstract and keywords. It is not intended as a judgement on the research; merely a count of the kind of word we would expect to indicate Phenomenon, Concepts, Theoretical framing, Empirical techniques, Analytical techniques. If all five are present, then a code of “5 PCTEA” will indicate this. If you feel the coding for this record is questionable, we welcome discussion around the terms we matched or the way we categorized them. The facet you would expect may not be coded, or a facet may be coded inappropriately. This can also bear on a larger question, of which facets should be treated as defining in construction management research. Please get in touch, and we will look at it. More details here