This study aims to address the labor-intensive process of bank transaction and invoice reconciliation by presenting a comprehensive framework that leverages multi-criteria optimization and neural networks. The literature reveals a significant gap in automating this process with high accuracy and efficiency. The method presented in this paper integrates optimization techniques and neural networks to minimize manual intervention and enhance the precision of the reconciliation process. Utilizing utility functions for multi-criteria decision-making can effectively combine optimization techniques to balance accuracy, processing time, and error rates. Findings indicate that the proposed framework significantly improves processing time and accuracy, providing a scalable and efficient solution for financial operations. The presented framework demonstrates its potential by achieving a high level of F1-score, reducing processing times, and minimizing error rates, thereby advancing the automation of financial reconciliations.