The algorithm is adaptive so it may learn from your new data. This means if new intents appear along time, they will be discovered automatically without the user having to manually look for them which is very useful as text data is very dynamic.

You can trigger this process by manually uploading a new dataset (or using the API) into a project. This will create a new classifier were you can check what intents are new and which one changed (they have a new name and model based on the new information).

Did this answer your question?