End-to-end Procedural Knowledge Management in Industry 5.0 with AI-in-the-loop

Presentation of the In Use paper at EKAW 2026

Authors: Mario Scrocca, Valentina Anita Carriero,  Laura Waltersdorfer, Gregor Käfer, Umutcan Serles, Ioan Toma and Irene Celino

Abstract: Industry 5.0 shifts the focus from technology-driven efficiency to a human-centric approach that leverages data, AI and robotics for optimised and sustainable operations. This transformation places new demands on how industrial know-how is managed. Capturing and exploiting procedural knowledge is one of the challenges that organisations face as people rely on fragmented sources (expert experience, legacy documents, and partial execution logs) with little standardisation and governance. To address these challenges, we identify the core activities for end-to-end procedural knowledge management, from the extraction and curation of procedures to the guidance of their execution and the enabling of auditing operations. The proposed activities are designed to augment operations by putting AI-in-the-loop while keeping the human in control. A system has been implemented to validate the approach by leveraging Knowledge Graphs to guarantee semantic interoperability and AI grounding.
We validated it in three industrial use cases: safety procedures in manufacturing, CNC machine commissioning, and renewable energy microgrid management. In each scenario, the system captures domain-specific procedures into a unified knowledge representation, enabling users to retrieve precise, context-aware procedural guidance. 
We present the results of the pilot evaluation with stakeholders and end-users at each company, which demonstrated improved procedure quality and execution efficiency, along with the lessons learned.