By using the State ADT to define how our application state transitions over time, we clear up the need for a some of the boilerplate that we typically need for a solid Redux integration. We can keep our action names and creators in the same file as the reducer they are used in, as we no longer need to have separate files to hand the same action across different portions of our state.
We will build our a single reducer that will handle the integration with Redux and organize our State ADT based reducers in their own files and bring them into our main reducer to handle running our state transitions and the cases where we do not have an transition for a given action.