Abstract

Organizations are beginning to distribute cognition, decision preparation, execution, monitoring, and learning across humans and AI agents. This shift is not only a problem of AI adoption, automation, or digital maturity. It changes the assumptions under which organizational actors are identified, authority is allocated, delegated action is attributed, adaptation is governed, and organizational continuity is preserved. Existing literatures on organization design, enterprise architecture, dynamic capabilities, sociotechnical systems, AI governance, and multi-agent coordination each address parts of this problem, but they do not yet provide an integrated theory of legitimate, accountable, and adaptive mixed human-AI organizing. This paper develops Constitutional Organizational Architecture as a higher-order, configurational construct explaining how organizations constitute and govern distributed human-AI action. The theory argues that adaptive enterprises require enduring but revisable constitutional foundations that specify authority provenance, accountable delegation, bounded autonomy, human-AI role differentiation, representational correspondence, governance observability, adaptation, constitutional evolution, and resilience and repair. Twelve propositions explain how these elements operate through authority-legitimacy, delegation-accountability, constraint-enabled autonomy, correspondence-observability, topology-adaptation, continuity-adaptation, representation-learning, meaningful-human-governance, and resilience-repair mechanisms. The paper contributes by reframing AI-native organization as a constitutional organizing problem, developing Constitutional Organizational Architecture as a focal construct, and advancing an authority-centered theory of AI governance that distinguishes technical autonomy from organizational authority. It offers a research agenda for empirical validation without claiming empirical evidence.