Foundational Concepts

The theory requires a small set of conceptual distinctions. These distinctions are not a glossary; they define the terms needed to make the propositions coherent.

An AI-native organization is an organization in which humans and AI agents participate in cognition, decision preparation, execution, governance support, and learning. The term does not mean that AI replaces humans, that all work is automated, or that the organization is technologically mature. It means that AI participation is sufficiently embedded that the organization must design for mixed human-AI action.

An adaptive enterprise is an organization capable of changing roles, processes, representations, capabilities, and topology while preserving identity, authority, accountability, and continuity. Adaptation is not merely speed; it is governed change under constraints.

An organizational constitution is an enduring but revisable source of principles, authority rules, role rules, representation requirements, governance commitments, adaptation procedures, amendment pathways, and repair mechanisms. It is not necessarily a single legal document. It may be represented across formal charters, policies, operating principles, decision rights, architecture standards, and governance records. Its theoretical role is constitutive: it specifies the conditions under which organizational action is recognized as legitimate.

The focal construct of the paper is Constitutional Organizational Architecture. The terms that follow play different theoretical roles. Some are dimensions of the focal construct, some are mechanisms explaining its effects, some are outcomes, and some are boundary conditions or failure modes. The theory does not claim that every term is an independent construct of equal status.

An organizational actor is a human, collective, or AI participant that can occupy a governed role in organizational action. A human actor may hold judgment, authority, and accountability. An AI agent is a non-human computational actor that may act within delegated, supervised, and revocable scope. An AI agent may execute action, generate representations, or support decisions, but it is not treated here as a moral or institutional authority source. A collective actor is a team, unit, committee, or agent collective participating under a governed role or jurisdiction.

An identity identifies an actor or collective as a recognized participant. A role defines expected participation, obligations, permissions, and relationships. A role is not the same as an actor: actors may enter, leave, or change roles, and AI agents may occupy only roles for which they are authorized and supervised.

Authority is the legitimate organizational capacity to make, delegate, constrain, or approve action. Permission is a specific allowed action under authority and policy. This distinction is central. A system may have technical permission to invoke an API or update a record without having legitimate organizational authority to make a decision. Jurisdiction defines the domain within which authority applies. Delegation transfers bounded action scope from an authority holder to a role, actor, or AI agent. Delegation does not abandon accountability. Accountability is traceable responsibility anchored in human or organizational authority and governance.

Bounded autonomy is the freedom to act within delegated scope, constraints, supervision, and revocation. Autonomy is not authority; it is discretion under authority. Meaningful human governance is the capacity for humans or human-governed bodies to understand, contest, intervene, override, amend, or repair high-impact or ambiguous AI participation. Nominal human review is insufficient if humans lack the information, time, authority, or competence to govern. Meaningful human governance therefore requires evidence access, interpretive competence, decision authority, practical time to intervene, contestability, escalation pathways, and accountability for the governance act itself.

Organizational topology is the dynamic pattern of roles, authority, information, dependencies, modules, and relationships. It is broader than an organizational chart because it includes temporary configurations, interfaces, cross-unit dependencies, AI-agent participation, and representation flows. Organizational runtime is the enactment of roles, decisions, workflows, representations, AI participation, governance observations, and learning under constitutional constraints. The term is organizational, not software technical.

Representational correspondence is the preservation of meaningful relationships among organizational representations, such as purpose, authority, decisions, execution, outcomes, and learning. It is not identical to factual truth or documentation completeness. Governance observability is the capacity to see and interpret whether action corresponds with constitutional commitments. More data does not necessarily create more observability.

Finally, adaptation refers to runtime or organizational adjustment within constitutional authority, while constitutional evolution refers to governed change to constitutional principles, authority rules, or amendment basis. Learning may trigger proposals for constitutional evolution, but it should not silently alter constitutional commitments.