Organizational Design Meets Agentic AI: Why Multi-Agent Systems Need Management Theory

Abstract: While individual AI capabilities and limitations—the "jagged frontier"—are increasingly documented, multi-agent AI systems introduce organizational-level complexities that lack established frameworks or vocabulary. Current approaches to agentic workflows draw heavily from software engineering paradigms (control planes, orchestration loops, API hooks), but these technical metaphors inadequately address coordination failures, authority ambiguities, and emergent dysfunctions familiar to organizational scholars. This article argues that management theory—spanning boundary objects, spans of control, decision rights allocation, and organizational architecture—offers essential conceptual tools for designing and governing multi-agent systems. By integrating organizational design principles with technical implementation practices, practitioners can move agentic AI from experimental art toward evidence-based organizational capability. The synthesis identifies parallels between classic organizational pathologies and observed multi-agent failure modes, proposes a management-informed vocabulary for agentic systems, and outlines evidence-based design principles that balance automation efficiency with human oversight, structural clarity, and adaptive learning.

See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.