<p>Under the “One Country, Two Systems, Three Legal Jurisdictions” framework, cross-regional collaboration in the construction industry of the Guangdong–Hong Kong–Macao Greater Bay Area (GBA) is more complex than other urban agglomerations, and this makes traditional methods relying on manual analysis of influencing factors inefficient and subjective. While existing large language models (LLMs) can meet the needs of intelligent applications, they lack specific domain knowledge. Considering the intelligent advantages of LLMs, this research proposes a lightweight expert agent through providing an optimised low-rank adaptation (LoRA) model SVDSR-LoRA, integrating domain knowledge using the fine-tuning method. Experiments on the 1.5b lightweight base model of qwen2.5 and deepseek-r1 show that the proposed SVDSR-LoRA training method can increase the mid-term convergence speed by 36%–50% compared with the standard LoRA method. A lightweight multi-expert agent influencing factor analysis system is constructed to simulate a collaborative analysis environment with experiments showing the hit rate in high-frequency influencing factors reached 100%. The comprehensive evaluation scored at 4.03/5.0 and demonstrated the capability in revealing the external environmental factors influencing cross-regional collaboration in the construction industry of the GBA systematically, which was significantly better than that of a single model (2.52–3.22/5.0). The proposed SVDSR-LoRA training model and the MAS system establishing pattern can provide a reference for rapid and effective lightweight agent system construction methods for intelligent analysis tasks of cross-regional collaboration in the construction industry of the GBA, as well as other similar LLM-based cross-regional collaboration research</p>