Relato de experiência no uso de engenharia de prompt para apoio ao desk review editorial

This study aims to propose a set of prompt engineering tactics to be applied with generative artificial intelligence as an assistant in the desk review process of scientific journals. Methodologically, the research is classified as exploratory, bibliographic, and descriptive. The results show that AI was efficient in simple tasks by generating quick and clear analyses, but it failed in more precise quantitative verifications. Finally, it's concluded that AI agents are useful as initial support but do not replace human review. It is hoped that this study will encourage new tests in editorial practices within scientific communication using generative artificial intelligence.