Automação de verificações manuais em processos de KYC.

Abstract

In recent years, sports betting has experienced rapid growth due to the digitalization and accessibility of online platforms. However, this growth has also generated challenges, such as market integrity, combating organized crime practices such as money laundering, and consu-mer protection. In this scenario, document verification and validation processes, known as KYC (Know Your Customer), have become essential to ensure regulatory compliance and user security. This study focuses on the challenges faced by bookmakers when implementing KYC processes in compliance with regulations, especially the need for manual analysis to validate KYC verification processes. Such manual verifications are typically time-consuming and costly. Given this scenario, an automation tool was developed that uses a LLM (Large Language Model) model to reduce the dependence on human intervention and improve the efficiency of the KYC document verification process. The methodology used covered a sample of 163 user data from the period December 26, 2023 to February 26, 2024, taking into account the main failures identified during this time interval. The results obtained show a significant improvement in the performance of manual analysis checks. Thus, of the sample carried out with 163 cases, the algorithm provided a verdict for 73.6% of cases, with 81.05% accuracy in correctly categorizing the user's analysis status. This highlights the overall effectiveness of the algorithm, making the process more agile and efficient, significantly reducing the human workload. On the other hand, 26.4% of the cases still required manual analysis for greater accuracy, highlighting situations in which the algorithm was unable to make an automatic decision. However, the research not only offers a practical and technological solution to the challenges faced by bookmakers, but also contributes to the advancement of the automation of document verification processes in regulated environments.