Decision Architecture: Integrating the Third Axis into Systems Engineering in the AI Era

This paper proposes Decision Architecture (DA) as a third independent axis that complements System Architecture (SA: spatial/structural dimension) and Mission Architecture (MA: temporal/purposive dimension) in systems engineering and enterprise architecture practice. Conventional architecture frameworks address what to build (SA) and when/why to execute (MA); however, who decides and how decisions are made are typically left implicit or treated as secondary concerns. This gap becomes critical in complex socio-technical systems and increasingly in AI-era contexts, where explicit decision-making structures are essential for accountability, governance, and human-AI collaboration. We introduce CIO-GoF (ContextIntelligenceOrchestration with Governance of Feedback) as DA's core structural model. This framework demonstrates cross-domain applicability across diverse systems engineering contexts and industries, providing a common conceptual language for decision-making architecture in complex systems. Our primary conceptual contributions are threefold. First, we identify three analytically separable architectural questionsWhat (SA), When (MA), and Who (DA)each representing an independent dimension of system design. Second, we distinguish three distinct qualities of Why and How across these axes: Static (SA), Dynamic (MA), and Orchestral (DA). Third, we demonstrate that DA possesses a fundamental duality: it functions as a governance mechanism enabling intentional design (the Design Perspective), while simultaneously manifesting as observable decision-making patterns derived from 5W1H analysis (the Analytical Perspective). This dual-perspective framework provides practical design principles for governance structures and establishes theoretical foundations for ensuring human oversight in safety-critical and regulated decision-making contexts. Cross-domain examination, including a detailed case study of civil aviation type certication, demonstrates DA's applicability across diverse temporal scales and governance contexts relevant to systems engineering practitioners.