Agentic Manufacturing Operations Management: Concept and First Prototype

Manufacturing Operations Management (MOM) systems, based on ISA-95, dominate industry practice, thanks to their centralised, deterministic control and mature standards. In contrast, multi-agent manufacturing systems promise self-organising, resilient, and highly flexible production, yet remain largely confined to laboratories as their distributed intelligence is challenging to engineer, govern, and certify. This paper introduces Agentic Manufacturing Operations Management, a hybrid concept that fuses the strengths of both paradigms. The eight core MOM activities, product definition, resource management, scheduling, dispatching, execution management, data collection, tracking and performance analysis are reinterpreted as autonomous agents that coexist with traditional MOM services. The proposed conceptual architecture includes an agent layer aligned with ISA-95, a digital twin layer providing semantic consistency, and the legacy MOM systems layer as a transactional backbone. As an exemplary instantiation, a prototype in which large language-enabled agents support the design of experiments process in slurry mixing for battery production is implemented. The prototype demonstrates how agents can assist in defining, executing, and analysing both full and fractional factorial designs, reducing manual effort.