Logic hybridization in enterprise platforms: Reconfiguring architecture and governance for generative AI

Enterprise platforms have historically relied on a logic of determinacy to ensure predictable, reproducible, and auditable system behavior through mutually reinforcing architecture and governance mechanisms. However, generative AI (GenAI) introduces a conflicting logic of probabilistic inference that destabilizes these mechanisms by generating variable, context-dependent, and non-deterministic outputs. As platform owners increasingly integrate GenAI into enterprise platforms, this study draws on assemblage theory and an in-depth case study of Salesforce to examine how enterprise platform owners reconfigure architecture and governance to manage the coexistence of deterministic and probabilistic logics. We develop a process model of logic hybridization that explains how platform owners restore system-level stability while accommodating and regulating probabilistic components. Specifically, we identify four complementary mediation mechanisms, that is, dependency mediation, service mediation, interaction-centric control, and dynamic certification and monitoring, through which platform owners reconfigure architectural and governance arrangements. We further theorize that logic hybridization gives rise to a hybrid platform logic characterized by three emergent properties: boundary fluidity, trust mediation, and expanded generativity. This study contributes to research on enterprise platforms and enterprise system evolution by identifying incompatible system behavior as a new source of platform destabilization, extending theories of platform evolution beyond structural change, and theorizing how deterministic and probabilistic logics coexist under mediated conditions in enterprise platforms.