Abstract for: Integrating System Dynamics Principles into Large Language Model-Assisted CEO Decision Support: A Framework for Strategic Transition Management

Between 40-70% of strategic initiatives fail, particularly in AI transformations and mergers, due to executive reliance on LESR thinking—Linear, Episodic, Static, and Reductionist frameworks that overlook feedback loops, delays, and systemic interdependencies shaping organizational outcomes This paper synthesizes System Dynamics theory, dynamic capabilities literature, and LLM applications to develop a Digital Twin-based strategic architecture. It proposes institutional mechanisms including Friday Reflection practices and a Thriving & Flourishing Dashboard to embed systemic thinking into executive decision-making routines The SD-plus-LLM framework reveals counterintuitive dynamics in AI-merger contexts: productivity mirages from rapid automation, capability erosion through talent attrition, and worse-before-better transitions. The architecture enables executives to distinguish healthy learning dips from structural rejection, quantifying technical and cultural debt risks previously obscured by linear planning Integrating SD and LLMs transforms strategy from static planning to dynamic transition management. Success requires CEO upskilling through experiential learning, dedicated staffer support teams, and institutionalized feedback practices. The framework addresses Teece's dynamic capabilities requirements—sensing, seizing, and transforming—while overcoming Sterman-documented misperceptions of feedback that systematically undermine executive decision-making Edit, referencing