Settlement is an important phenomenon in geotechnical engineering for preventing structural problems in buildings. In order to quantify it, a property called compression index is required, which can be obtained through laboratory tests. However, these procedures can be costly in terms of time and equipment. In the last decades, literature has presented empirical equations and machine learning models for estimating soil properties. Furthermore, automated machine learning has emerged as an alternative for finding efficient models with simpler procedures. This study aims to develop AutoML models to estimate soil compression indices. To this end, two databases were selected, preprocessed and applied to three models: AutoGluon, FLAML and H2O. The performance metrics used were R2, RMSE and execution time. FLAML was found to have the most satisfactory R2 and RMSE values, while H2O had a considerably shorter execution time. It was concluded that the three models are suitable for the tasks, depending on the user's priority.