Facility management (FM) could be radically transformed for optimal efficiency by integrating generative artificial intelligence (AI) and digital twins (DT). However, such integration has not yet been explored by prior studies to provide theoretical clarity and address the fragmented knowledge to manage the FM system complexity. This study aims to propose an integrative framework combining DT and generative AI for FM in the built environment.
An expert-informed systematic literature review was adopted, including the review of 779 articles and validation via expert opinions. This is further integrated with content analysis.
Generative AI and DT were revealed to have been in existence since 2017 and had gained exponential momentum up to 2025, with a growth expectation beyond. Based on the synergistic relationship between generative AI and DT, the study developed generative digital twins (GenDT) for FM along with its socio-technical implementation framework, highlighting the social, technical, socio-technical and environmental dimensions to effectively implement it. Balancing GenDT’s strengths, weaknesses, opportunities and threats (SWOT) via a SWOT matrix analysis framework could promote effective FM performance. Five research directions were identified, namely, specific generative AI models for GenDT for FM; impact of system failures on bidirectional mapping and real-time performance; articulative and interactive relationship within the socio-technical framework for GenDT implementation; pilot testing of the GenDT implementation framework; and refining FM organisational functions for GenDT deployment.
The study introduces a novel GenDT paradigm for FM, integrating generative capabilities with socio-technical systems theory to capture the dynamic interplay between technology and organisational practices. It advances existing DT research beyond static models by enabling adaptive, human-centric FM processes. The proposed implementation framework operationalises GenDT in real-world contexts, addressing a key implementation gap in translating DT capabilities into adaptive, human-centric FM operations within the built environment.
