Despite the excellent performance of brain-computer interfaces (BCIs) based on code-modulated visual evoked potentials (c-VEP), these systems are inherently synchronous, resulting in constant selections even when the user is not paying attention to the visual stimuli. This significantly limits their application in real-world environments. The objective of this pilot study is to propose a multi-window correlation-based approach to implement both asynchronous and early stopping stages in real-time. In our system, a decision is made only when the system is confident that (1) the user is receiving visual stimuli and (2) the user is paying attention to a specific command. This approach was assessed offline with eight healthy users, achieving 100% accuracy in asynchronous detection and a mean decoding accuracy of approximately 95% with only 1.8 s per trial. These results demonstrate the effectiveness of the proposed method in monitoring user attention while maintaining competitive performance in c-VEP-based BCIs.