Let's reduce this to mechanisms: predictive analysis of case outcomes is not a simple subcategory of legal process automation.
The causal mechanism for predicting outcomes relies on structured data inputs and learning algorithms that identify patterns in judicial decisions.
In contrast, process automation streamlines repetitive tasks via rules and workflows.
A firm can very well invest in systems to predict the outcome of a case before the Council of State without automating a single document management process, because the constraints and objectives are distinct.
These are two applications of AI with sufficient conditions and fundamentally different operational goals.