The California Bearing Ratio (CBR) is a key indicator used to gauge the strength and condition of foundational soils. Yet laboratory tests to determine CBR values are often time-consuming and labor intensive. This research introduces a novel approach utilizing Neuroevolution of Augmenting Topologies (NEAT) to estimate the CBR of foundational soil. A dataset comprising 211 samples from existing studies was partitioned into a 70/30 ratio for training and testing. The model, based on regression, uses input variables such as the liquid limit (LL), maximum dry density (MDD), plastic limit (PL), and optimum moisture content (OMC), with the soaked CBR value serving as the output variable. Pearson correlation analysis revealed that LL, PL, and OMC are negatively correlated, while MDD is positively correlated with soaked CBR, with correlation coefficients −0.6, −0.41, and −0.83, and 0.85, respectively. The model’s performance is assessed with the root mean square error (RMSE), coefficient of determination (R2), and mean absolute error (MAE). The NEAT models demonstrate a simple structure and strong generalizability, achieving an RMSE of 9.477, R2 of 0.83, WI of 0.911, and MAE of 5.871. Inference is that the NEAT approach provides an efficient and reliable alternative to traditional laboratory methods for predicting CBR values. While an R2 value of 0.83 indicates a moderately strong correlation, the MAE of 5.871 and RMSE of 9.477 suggest there is room for improvement. However, the model’s overall performance metrics demonstrate its potential applicability in geotechnical engineering.