Year: 2026
Type: article
Abstract: The rapid evolution of Agentic AI offers unprecedented potential for automating complex business processes.However, organizations lack systematic approaches to transition from rule-based process models (e.g., BPMN) to autonomous Multi-Agent System (MAS) architectures, currently relying on adhoc trial-and-error.This Design Science Research (DSR) project proposes an intelligent translation framework to bridge this gap.Extracting structural and semantic characteristics from BPMN models, the framework utilizes a rigorous decision logic to methodically derive optimal MAS architectures, mapping process steps to suitable agentic behaviours and capabilities.Currently in the artifact design phase, this research aims to shift the paradigm from experimental AI tinkering toward structured AI engineering.At the TREO forum, we seek to discuss core challenges regarding the systematic mapping taxonomy, the evaluation dilemma of autonomous systems, and the future role of traditional process modeling.
Authors: Sardogan, Abdullah, Wirsam, Jan, Alt, Rainer
Language: English
Subfield: Management Information Systems
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7163840128