Inovação e gestão no poder judiciário: uma abordagem empírica com Inteligência Artificial Generativa para otimização e gestão de riscos de processos

The Brazilian Judiciary seeks to modernize its management to increase efficiency, facing barriers of structural rigidity and compliance. This study aims to develop and validate a technological artifact - a software tool based on Generative Artificial Intelligence (GAI) - to automate the analysis of work processes, aiming at workflow optimization and risk identification. The methodology employed is Design Science Research (DSR), structured into the stages of problem identification, objective definition, design and development, demonstration, evaluation, and communication. The core mechanism involves the transformation of BPMN mappings and Standard Operating Procedures (SOPs) into structured data (XPDL), processed by the Gemini 2.5 Flash model. The results, obtained through a simulated case study at the TRE-SP, demonstrated the tool's feasibility in generating optimization suggestions and identifying risks with precision comparable or superior to that of specialists. The work contributes a replicable software artifact (technical) and the advancement of knowledge concerning AI governance in the public sector, operating in strict compliance with CNJ Resolution No. 615/2025.