I have been using Rasa stack for chatbot development for the past three years. I have developed multiple chatbot systems using Rasa.

Rasa is probably the best python framework for building NLP based chatbot system. The machine learning models that has been used for intent classification, entity recognition and NLG models are highly accurate. The data format is very easy to interpret from Rasa.

There are no any features that I do not like about Rasa.

Very good experience using rasa

Rasa is a great tool for building AI powered chatbots. It's intent classification models are very accurate and named entity recognition models are also highly accurate.

Integrating rasa chatbots with other comunication channels documentation should be improved

Easy to develop the chat bots for clients and personal use. Can trigger Robocorp bots from the chat bots. This is open-source platform, which makes it more popular in building chat bots.

For now, there is nothing that makes me dislike this tool.

Es una herramienta muy potente y cuenta con gran cantidad de características que igualan a las soluciones de la competencia. Lo más importante de todo es que no cuenta con limitación de ningún tipo para contruir soluciones de asistentes viruales, para comercializar. Tampoco cuenta con tantas complicaciones para implementar en servidores con distintos sistemas operativos.

Tecnología de código abierto y gratuito. Dispone de documentación en constante actualización y crecimiento. Cuenta con gran cantidad de material para aprender a utilizar la herramienta. El foro oficial cuenta con un gran número de miembros que participan de manera activa.

No cuenta con soporte oficial con el idioma español, pero se lo puede implementar con pocos pasos haciendo uso de librerías externas, como spaCy.

Very good experience using Rasa Stack

I had used Rasa to develop a chatbot system for our bank and i found it the most suitable tool for building chatbot systems. The way Rasa takes in input data for the chatbot and the training time for such chatbots being really low is awesome.

I found debugging in Rasa difficult die to action server and rasa server running separetely on different servers.

I recently used Rasa to build a chatbot integrated with a FAQ.

I believe it is an excellent python framework. It is very simple to build both the NLU model and to configure the pipeline to use extremely powerful models of preprocessing and classification of intentions and entities. I also found it very intuitive to declare everything in the domain and work with stories and rules. Other structures such as slots and forms are also very practical and make all the difference when building the bot.

I liked all the features I found in RASA and they all helped in the development. I also thought the integration with some channels was easy.