Uma jornada de transformação digital no serviço público: rumo ao agêntico - Uma pesquisa focada na otimização de processos de trabalho

Abstract

This work presents research on the implementation of solutions using Artificial Intelligence in the context of the Brazilian public service, as a driving force for digital transformation. Based on the convergence of the high level of standardization of work processes, the availability of large volumes of data, accessible computing power, and the advancement of language models, we believe that the current moment is the ideal scenario for the proliferation of intelligent agents. Using an applied research format, functional prototypes were developed, including systems to support bidding, auditing, minutes generation, fact-checking, and specialized chatbots, with the aim of proving concepts. In addition, a classifier for synthetic texts was developed using fine-tuning techniques; databases were created and shared; and theoretical studies focused on the topic were conducted. The research demonstrates the technical feasibility and potential value of these systems, corroborating the idea that it is possible to develop effective agentic systems in accordance with the regulatory framework proposed by the CNJ (National Council of Justice). The results indicate significant gains in operational efficiency and service quality, maintaining a balance between agent autonomy and human supervision, constituting a powerful tool for optimizing administrative and judicial actions.