This paper provides a comprehensive analysis of the undrained failure envelope for embedded foundations in anisotropic clays. Using the AUS failure criterion as the soil strength model, the study examines how the anisotropic strength (re) and embedment depth (D/B) affect the behavior of the footing under combined loading conditions. Failure envelopes are assessed via two-dimensional finite element limit analysis (2D FELA) in both 2D and 3D spaces. This research highlights the failure mechanisms of embedded foundations, offering valuable insights into the engineering design of footings in anisotropic clays subjected to combined loads (V, H, M). Furthermore, this study introduces an advanced soft-computing approach by creating a machine learning model that leverages the adaptive neuro-fuzzy inference system (ANFIS) integrated with the particle swarm optimization (PSO) algorithm to predict the failure envelope of embedded footings, highlighting the novelty and original of this study. The optimised ANFIS model has been validated and demonstrates a strong correlation with the numerical FELA results, offering engineers a valuable tool for determining the failure envelope of embedded foundations in anisotropic clay under different loading scenarios (V, H, M).