From Historian to Agentic Intelligence: A Data Enablement Framework for Contextualised AI in ISA-88 Batch Manufacturing

Abstract

Modern batch manufacturing environments generate large volumes of operational data through industrial historians; however, much of this information remains underutilised because it lacks process context and standardised representation. This paper suggests a framework for the data enablement of historian data to provide contextualised intelligence for agentic artificial intelligence applications in ISA-88 batch manufacturing systems. This framework brings together the historian records and the concepts of ISA-88, such as recipes, procedures, equipment, process stages, and information about the execution of the batches, providing a structured and context-aware data foundation. The raw manufacturing data is preprocessed, contextually mapped, enriched with meaning, and represented in the form of knowledge for establishing meaningful relationships between process events and operational assets. The contextualised information is then fed into an agentic AI layer that can make autonomous decisions, offer decision support, detect anomalies, optimise processes and provide operational recommendations. By combining industrial data management principles with intelligent agent technologies, the proposed framework enhances traceability, improves decision accuracy, and supports real-time manufacturing intelligence. Evaluation measures include accuracy of context extraction, effectiveness of the decision, process efficiency, and response time. By supporting the scalability, explainability, and context-awareness of AI solutions in today's batch manufacturing environment, the framework paves the way for the advancement of Industry 4.0.