Artificial intelligence is rapidly evolving from isolated prediction and recommendation systems into institutional infrastructure that increasingly influences enterprise workflows, operational decisions, governance processes, software engineering, cybersecurity operations, and business execution. Yet many organizations continue to approach AI primarily as a model-selection problem rather than a representation, governance, and institutional-design challenge. This paper introduces SENSE-CORE-DRIVER as a governance architecture for enterprise AI systems. The framework argues that AI systems do not operate directly on reality, but on machine-legible representations of reality. SENSE represents the legibility layer where signals, entities, states, and evolution transform fragmented institutional reality into structured machine-readable understanding. CORE represents the cognition layer where models, reasoning systems, planning engines, orchestration logic, and optimization mechanisms interpret that reality. DRIVER represents the legitimacy and execution layer where delegation, representation, identity, verification, execution, and recourse determine whether AI-driven action is authorized, accountable, and correctable. The paper argues that many enterprise AI failures emerge not merely from weak models, but from weak representation infrastructure, unclear authority boundaries, fragmented context, insufficient runtime governance, and inadequate recourse mechanisms. It introduces key concepts including representation integrity, governed execution, runtime legitimacy, autonomy allocation, and three structural tensions in enterprise AI governance: the representation-legitimacy tension, the human-oversight tension, and the simulation-reality tension. Positioned at the intersection of enterprise architecture, AI governance, institutional systems, and machine-legible reality, SENSE-CORE-DRIVER provides a practical conceptual framework for CIOs, CTOs, enterprise architects, governance leaders, researchers, and policymakers seeking to design scalable, trustworthy, and operationally legitimate enterprise AI systems. Author: Raktim SinghWebsite: https://www.raktimsingh.comORCID: https://orcid.org/0009-0002-6207-602XGitHub Repository: https://github.com/raktims2210-dev/representation-economyFigshare DOI: https://doi.org/10.6084/m9.figshare.32393949
RAKTIM SINGH (Sun,) studied this question.