This study presents Smart Audit in Real Time, a mobile system designed to automate the visual inspection of technological infrastructures using OpenAI GPT-4o Vision. The application captures images, identifies auditable elements, and generates structured reports aligned with international IT auditing standards. A dataset of 50 images from different operational environments was analysed, producing 207 findings grouped into four categories: physical security, ergonomics, environmental conditions, and regulatory compliance. The system achieved an overall accuracy of 83% and an average F1-score of 0.65, with a mean response time of 15.8 seconds per evaluation. A chi-square test applied to the confusion matrix confirmed a statistically significant association between predicted and actual categories, supporting the reliability of the classification. The results show that automated visual auditing from mobile devices can efficiently detect operational risks and produce consistent evaluative outputs without expert intervention. This research contributes to the digital transformation of IT auditing by improving accessibility, precision, and real-time decision-making.
Huamanchumo-Trujillo et al. (Mon,) studied this question.