The AI model learns through training. Training is when you assign a customer message, or expression, to a specific customer query, or intent.
Through training, the AI model learns patterns in the data so that it can distinguish individual intents from one another. As you train, the AI model learns and makes adjustments, so that its predictions are more accurate.
In the Dashboard, under AI > Models, you will see that the AI automatically makes adjustments every evening and creates a new model, if any changes have been made to the training data that day. If you have added new intents or have trained an intent with new expressions and you need the AI model to learn immediately, select “Train Model”.
An evaluation model is also automatically created once a week, whether training has taken place or not. This will generate an updated Confusion Matrix available under Training Center > Confusion Matrix.