A New Paradigm of Business Computing Driven by Domain-specific Foundation Models

With the accelerating process of enterprise intelligence，business computing （i.e.，an important branch of computational social science） has evolved from traditional business analytics into intelligent systems integrating big data and artificial intelligence. However，current practices remain constrained by the heterogeneity and dynamism of multi-source data，insufficient model interpretability，and the complexity of system deployment，which limit its broad application in complex business environments. To address these challenges，this paper proposes a domain-specific foundation model-driven technical architecture as a potential breakthrough. It constructs a four-layer research framework encompassing the data layer，system layer，business layer，and governance layer，systematically outlining the evolution of business computing. The framework demonstrates advantages in effectively integrating multi-source heterogeneous data，enhancing model transparency and dynamic adaptability，and reducing computational costs，thereby improving the adaptability and robustness of business computing systems across diverse operational contexts. Building on this framework，the study examines two representative application contexts （i.e.，marketing computing and organizational computing） to systematically analyze their core scientific issues and showcase the value of domain-specific foundation models in addressing external market dynamics and internal management complexities. Finally，it outlines key future research directions for business computing，including human-AI collaboration mechanisms，intelligent integration pathways，information updating strategies，and ecosystem dependencies，aiming to provide theoretical guidance and methodological support for enterprise intelligent decision-making.