Knowledge integration for crisis management: the mediating mechanism of transactive memory systems and how AI disrupts it

Purpose This study aims to investigate the mechanisms through which knowledge integration (KI) depth enhances corporate crisis management (CCM) efficacy. Moving beyond a direct-effect model, it examines the mediating role of transactive memory systems (TMS) and the critical negative moderating effect of artificial intelligence (AI) counter-productivity on the KI–TMS relationship. Design/methodology/approach A sequential explanatory mixed-methods design was used. Study 1 used a quantitative cross-sectional survey of 492 knowledge workers to test the hypothesized relationships (H1–H3). Study 2 then undertook a qualitative, 14-month ethnographic multi-case study of four organizations, collecting data from 62 interviews, 42 observations and 215 documents to explicate the underlying “mechanism-evidence” chain. This design allows the qualitative findings to explain and contextualize the initial quantitative results. Findings The quantitative results confirm the hypothesized model: KI depth positively influences CCM efficacy (H1), TMS mediates this relationship (H2) and AI counter-productivity negatively moderates the KI–TMS link (H3). The qualitative analysis elucidates how this occurs, revealing that TMS functions as a three-stage sequential mediator (specialization, credibility and coordination) to translate KI into action. It further specifies that AI counter-productivity disrupts this precise pathway: information overload undermines specialization, relational alienation erodes credibility and cognitive inertia impairs coordination. The integrated findings highlight that this disruption is not inevitable but contingent on organizational design. Originality/value First, it specifies the sequential socio-cognitive mechanism by which AI counter-productivity disrupts knowledge processes, moving beyond generic stressor models. Second, it questions assumptions of technological determinism by illustrating a shift from “technological fix” to “organizational immunity,” suggesting resilience can stem from human-centric governance that preemptively neutralizes technological disruptions. Third, it proposes the functional equivalence of social capital, as evidence indicates rich relational networks can compensate for a lack of technological advancement in enabling effective TMS, offering a “low-tech resilience” pathway. Fourth, it reconceptualizes TMS as a dynamic, enactable capability, a “living protocol” activated under crisis, providing a micro-foundation for understanding knowledge mobilization under duress. Collectively, these contributions offer a process-based theory of how KI translates into CCM efficacy through mechanisms that are vulnerable to, yet defensible against, technological disruption.