Integrating human capabilities and needs while accommodating dissonances caused by human-machine interactions and the emergent behavior of intelligent devices is a critical challenge in Cyber-Physical Production and Human Systems (CPPHS). This paper proposes a Multi-Agent CPPHS approach that addresses the lack of appropriate abstractions for interaction, coordination, and synchronization between human and artificial entities. The virtual representation of physical entities is achieved through a set of specialized agents, enabling complex entity representation. Moreover, a modified potential field approach (PFA) is employed as the interaction protocol, ensuring efficient and effective cooperation between artificial and human entities by incorporating the human operator's fatigue level. The proposed approach was evaluated through two case studies: a discrete-event simulation based on related literature and a realistic human-robot assembly system. Results demonstrate the agents' ability to maintain updated information, make decisions based on entity state (i.e., fatigue level), and achieve workload balance. Furthermore, the modular entity structure exhibited flexibility and adaptability to different entities and conditions. This work paves the way for enhanced human-machine collaboration in CPPHS by providing a framework that seamlessly integrates human and artificial entities while accounting for their unique characteristics and requirements.