ML models are very dependent upon the datasets used to train and test the given model. A frequent problem in ML is overfitting. Overfitting means that the model does not generalize well from training data to unseen data, especially data that is unlike the training data. Common causes include the presence of bias in the training data, meaning the model cannot distinguish between the signal and the noise.
AI Builder implements many techniques to avoid such problems, but you will need to supply AI Builder with enough data to be able to create a model. The more data and the more varied the data, the better the model will behave.
AI Builder requires the training and test data to be stored in entities in the Common Data Service. If the data does not reside in the Common Data Service, you will need to import the data. You may need to create a custom entity for this data.