Trustworthy Agentic Supply Chains: A Governance Framework for Digital Twin Orchestrated AI Decisioning Under Compliance, Auditability, and Data Sovereignty Constraints

The rapid increase of Artificial Intelligence (AI) within Supply Chain Management (SCM), has transitioned SCM from using primarily Predictive Analytics & Decision Support, toward increased Autonomy of Decision Execution. However, although there are many examples of AI-driven SCM systems currently being used, they generally suffer from low levels of Trust, poor Governance structures, inadequate Auditability, and unresolved Data Sovereignty issues; all of which limit their potential deployment in High Consequence & Regulated Operational Environments. In an effort to address this important gap, this research introduces a comprehensive Governance First Framework for Trustworthy Agentic Supply Chains; where Autonomous AI Agents use Digital Twin Orchestrated Decision Intelligence to make decisions in accordance with explicit Compliance, Auditability, and Sovereignty Constraints. Agentic Supply Chains are defined as Socio Technical Systems, where Decision Authority is delegated to AI Agents that Continuously Sense, Simulate, Decide, and Act Across Dynamic Supply Networks. Digital Twins are redefined from Passive Visualization Tools to Active Orchestration Substrates that facilitate Real Time State Synchronization, Policy Execution, and Controlled Interaction Between Autonomous Agents and Enterprise Systems. On top of this base, the paper provides a Layered Reference Architecture for integrating Agentic Decision Intelligence, Bounded Autonomy, Governance by Design, and Human Oversight into a Unified Operational Model. The Framework addresses Key Adoption Barriers via Explicit Mechanisms for Regulatory Alignment, Decision Traceability, Data Sovereignty Preservation, and Risk Containment. The Architectural Constructs provided include Agent Drift Detection, Rollback and Safe Recovery, Simulation Based Stress Testing, and Resilience under Adversarial and Extreme Disruption Scenarios. Through the embedding of Governance within the Decision Architecture, the proposed model allows Autonomous Supply Chain Systems to be Auditable, Compliant, and Strategically Controllable while Retaining Adaptive Intelligence. Additionally, beyond technical design, the paper outlines Evaluation Metrics, Organizational Integration Principles, Ethical Considerations, and Strategic Implications related to Delegating Decision Authority to Agentic AI Systems. Finally, the Study identifies a Forward-Looking Research Agenda addressing Multi-Agent Coordination, Cross-Enterprise Autonomy, and Next Generation Optimization Paradigms. Overall, this Work establishes Trustworthy Agentic Supply Chains as a Distinct and Necessary Evolution of Supply Chain Intelligence, providing a Reusable Reference Framework for Researchers, Practitioners, and Policymakers looking to operationalize Autonomous Decision Systems Responsibly and at Scale.