Machine learning can play an important role in making subsurface quantitative interpretation workflows more efficient, consistent and potentially more accurate. Two workflows are shown in 1D and 3D applications. It is argued that the 1D cases are more about improving efficiency whilst the 3D cases have the potential to improve the accuracy. Examples are shown from conventional and unconventional basins. Beyond that, it is demonstrated how one can combine deep learning and physics-based models to provide fast and accurate subsurface predictions.