This paper presents the challenges and advancements in wave energy converters (WECs), focusing on the TALOS WEC, which utilises machine learning to predict power output. While this model shows high short-term accuracy, its long-term predictions suffer from increasing errors. To address this, the paper proposes a dual-model approach, where a supplementary error prediction model is incorporated alongside the primary model to monitor and correct long-term prediction errors. This approach has shown promising results in improving the accuracy of long-term predictions, marking a significant step in WEC research and applications.