TWX Analytics Builder supports batch scoring, but if you want to score new data, it forces you to add that data to the dataset of the model you want to score on. There should be a choice if the user wants to add the scoring data to the existing data set or create a new one, because otherwise the dataset will grow inevitably and the original state can not be restored as there is no method to remove data from a dataset.
Also, when viewing the model, users will be led to think the model was trained on the now extended data set, as it is not clearly visible in the UI that the data set as of training time is different from the current data set (you will have to check row counts and used filters and so on to find out about that).
