Reforecasting the ultimate load-bearing capability of piles, attained by employing diverse experimental strategies with field data, emerges as a notably essential and intricate pursuit in the realm of pile analysis and construction. The study primarily aims to formulate innovative AI predictive schemes to project pile-bearing capacity (PBC). The predictive model employed in this study relies on the Naïve Bayes (NB) scheme. A novel hybridization strategy has been employed to attain optimal and precise predictive outcomes, combining the Cheetah optimizer (CO) and the Jellyfish Search Optimizer (JSO). A database comprises diverse traits of piles and soil qualities gathered from various sources in the literature, including data from CPT and outcomes of pile loading tests. These databases are deployed for the developed schemes’ training and testing phases. The strategy employed in this investigation yielded precise outcomes, underscoring the efficacy of the recommended schemes. The NBCO model performed very well, with an R2 value of 0.9911 and an RMSE value of 179.4563, which indicated the best result.