Crossing the Rubicon of AI use: how employees’ experiences, behaviors, and mindsets shape AI adoption at the workplace
Although private use of artificial intelligence (AI) is rapidly increasing, employees in large organizations often remain stuck in experimental or sporadic AI use at work. This exploratory study investigates how employees progress from initial AI intentions to frequent and routinized AI use by drawing on the Rubicon Model of Action Phases. Based on semistructured interviews with twelve mid- to senior-level practitioners from large organizations and a Gioia-style analysis with substantial inter-coder reliability (κ = .72), this study reconstructs trajectories of employees’ use of enterprise AI applications (e.g., internal assistants or copilots) across four Rubicon phases: pre-decisional exploration, pre-actional commitment, actional implementation, and post-actional evaluation. The findings map four phase-typical patterns of experiences, behaviors, and decision mindsets and identify three Rubicon-threshold mechanisms that enable employees to translate initial intentions into routinized use of enterprise AI applications: (1) breakthrough value on real work tasks, (2) organizational legitimacy through secure, governed, and workflow-integrated enterprise AI systems, and (3) stabilization via routines supported by calibrated trust and, for some, an “AI-savvy” professional identity. This study contributes a phasebased process account of workplace AI adoption, a mechanism-based explanation of why strong intentions often remain sporadic, and phase-specific intervention guidance for deploying, governing, and workflow-embedding enterprise AI systems to enable sustained, routinized use.