Feedback automatizado em modelagem BPMN: uma análise do desempenho de ferramentas de IA e ferramentas de modelagem no suporte a verificação de corretude de modelos de BPMN

Abstract

BPMN (Business Process Model and Notation) modeling is widely used for documenting organizational processes; however, due to its complexity, novice modelers often make frequent mistakes [1][9][10]. At the same time, advances in Artificial Intelligence technologies have enabled the automated analysis of process diagrams [4][7]. In this context, this study analyzes and compares tools that provide automated feedback for BPMN models. The tools ChatGPT, Gemini, DeepSeek, plz.ai, and FreeBPMNQuality were evaluated using the same structured prompt based on an academic assessment rubric. Each tool performed two independent analyses on two previously corrected BPMN models, allowing the evaluation of accuracy, consistency, and feedback quality. The results indicate distinct performance levels across the tools, depending on the evaluation criterion, and show that their use as support for human evaluation contributes to improving BPMN model analysis and correction.