KiCoPro – Improve and simplify process modeling with AI

AI-supported decision-making processes are playing an increasingly strategic role in industrial production. They are shifting decisions away from reactive, experience-based approaches toward data-driven, predictive, and partially autonomous systems. However, a prerequisite for productive, secure, and scalable use is the explainability of the models used. Only when decisions are traceable can companies integrate AI into their production processes sustainably.

DFKI's goal is to develop human-centered systems for dialogue-oriented interaction with explainable artificial intelligence (XAI) in industrial production. The focus is on the comprehensibility, transparency, and traceability of AI-supported decision-making processes in order to strengthen trust and acceptance in industrial practice. 

The core of the approach is an agent-based solution that enables users to interact with the underlying AI systems in natural language. They can ask questions, request explanations, analyze uncertainties, and understand or specifically adjust model decisions. The developed architecture not only makes AI systems in process planning and execution explainable but also actively involves them in a mutual dialogue. Decisions can be discussed transparently, alternatives can be compared, and models can be optimized iteratively.

The result is an interactive assistance system for decision support that uses AI in a comprehensible, controllable manner, thereby strengthening user confidence in the long term. The approach was developed in the EINHORN research project funded by the German Federal Ministry of Research, Technology and Space.