The User Responses added to the step above are processed using various Machine Learning algorithms to generate the models. These models are then used on the User’s every message to select the step which matches the requirements.
Once the user messages, the bot responses, entities, and integrations are added to the steps, you can TRAIN the bot.
Training helps the bot in -
- Understanding queries coming from the user
- Sending responses to the user
- Collecting information from the use
To generate these models, you will have to click on Train Bot. This will send an email with the result of the training process. It will also contain details of any probable errors which could come up because of incorrectly entered data.
Ideally, the bot takes around 30 seconds to a minute, to get trained.
For example, if you add very similar sentences in two different steps, then the resulting email will show details about the conflicting steps and responses.
Check the Training Suggestions email for conflicting responses or other errors in the User Responses in the bot.