Use the Rasa Platform interface to train the Rasa Stack as a team. Work together to get the best possible performance and see who added training data at which point in time.
Keep your models updated with the latest training data, which is automatically synchronised between the training UI and your backend.
You wouldn't ship code without tests, why should machine learning be any different? Monitor the performance of your system and run every new model through a number of test cases.
Raiffeisen, a leading Swiss retail bank, wanted to explore how to appeal to a younger audience. After the success of newcomer banks like N26 and personal finance bots like Cleo, traditional banks face fierce competition and have to improve their customer experience.
Raiffeisen has high quality human customer service operations and cares deeply about data privacy. In addition, most customer interactions are more complex than just a simple question and answer. They require multi-turn dialogue.
Using the Rasa Platform, Raiffeisen was able to develop a successful proof of concept that was able to handle longer dialogues and could be deployed on-premise. Their team followed the following process:
Further development of backend integrations and evaluation of production deployments.