We are planning to create a Product/Service which will allow customers to create bots and they will be able to embed these bots into their web pages.
As every bot will have its own intents and training data. Every bot will be train seperately but in order to answer the questions against a specific bot we will have to load the models for that bot and then process the query against.
Would not it be a problem for our product to scale in future when our clients/customer increase? Will we be able to scale it? Will Rasa be able to process the requests quicker?
According to my knowledge and whatever I have learned about rasa, It does not seem like a good choice to create such product. Please guide me.
I think there are several possible answers for your post and I’ll try to be as neutral as possible.
We are planning to create a Product/Service which will allow customers to create bots and they will be able to embed these bots into their web pages.
My company is doing something similar currently so I think I am indeed able to utter something to this case. Since we are quite successful with it, I’d be so free to say: Yes, you can achieve what you are asking for with Rasa.
Would not it be a problem for our product to scale in future when our clients/customer increase? Will we be able to scale it? Will Rasa be able to process the requests quicker?
Rasa can be used in several ways and don’t forget - it’s OpenSource - so basically, if something doesn’t fit your needs, then modify it. In terms of scalability, there are lots of possibilities. As stated out in one of my postings you are able to dockerize even complex architectures built with Rasa - like running several NLG-Servers/Action-Servers and additional software via docker-compose next to each other.
You are also free to decide to use OpenShift or CloudFoundry to deploy your bots wherever you want to and to achieve horizontal and vertical scalability… don’t forget the core functionality of Rasa. It is definitely not caring about your CI/CD Pipeline or remote deployment strategy.
So the answer to your question “Will we be able to scale it” is easy - it’s a counterquestion: “Will you?”.
According to my knowledge and whatever I have learned about rasa, It does not seem like a good choice to create such product. Please guide me.
As already mentioned I think you are mixing things up here. According to the main-page:
“Machine learning tools for developers to build, improve, and deploy contextual chatbots and assistants. Powered by open source.”
Rasa won’t create the product for you. Every product has requirements. I think you should decide whether Rasa fits those requirements. If you need help with that, we need to know the requirements.
Requirements are quiet simple. We are creating a product where user will be able to create bot. The bots on our platform will not have any complex functionalities, they will be able to answer simple questions which will be set by the user in the form of text.
Thanks for the great answer @JulianGerhard, what you said is pretty much our answer as well. @hassanrafay there shouldn’t be any problems with doing this