I want to know what dialogflow does so that your training is immediate and if it is possible to simulate the immediate training. It is that sometimes I only change very little of my bot and I have to train everything, I would like only the new change to be added and that it is not necessary to train the entire bot. Something like a training in parts.
With Rasa, you execute an explicit rasa train command either in your IDE, Rasa X or from a CI/CD pipeline. If there is no change in the training data, the train will be skipped. Otherwise a new model will be trained.