The adoption of Digital Twin (DT) technology and the subsequent evolution of DT requirements and functionality has led to the recent convergence of DTs and Multi-Agent Systems (MASs). This paper examines the integration of DTs and MASs to address the challenges and opportunities for intra-organizational decision making and inter-organizational collaboration. While DTs provide high-fidelity representations of physical assets, their utility is often constrained by organizational reluctance to share sensitive data. MASs, in contrast, offer a decentralized framework for negotiation and coordination, enabling organizations to collaborate through autonomous agents that safeguard proprietary information. We argue that maintaining a conceptual separation between DTs and MASs - where DTs serve as reliable narrators of system state and agents act as negotiators – can enable secure, adaptive collaboration between organizations. Through an illustrative example focussing on a circular economy supply chain, we envision a system architecture for DT-MAS integration and the opportunities that may arise from an implementation thereof.