Missing knowledge to go from Rasa to web app implementation

Can someone just give me a step by step guide how to I actually deploy Rasa somewhere?

For example: Rasa -> Docker -> AWS EC2 -> NodeJS -> SQL Lite? Is this proper view?

I’ve made some ChatBot locally with Rasa but I’m not sure how to proceed from there to actually implement it in coordination with AWS services.

Hi there, here are the instructions for running Rasa in Docker: Running Rasa with Docker

If you have deployment experience of how to deploy dockerized/containerized apps on an EC2 instance, then you should be fine – unfortunately we don’t have step-by-step deployment for that, because as you know there are tons of different services all with different setups. This is why we have docker as the guideline.

In a similar was, if you want to use NodeJS as your front-end (input channel) or backend (custom actions), you can do that by using our REST channel or python SDK as an example of how to create these with the api endpoints we provide.

Can you elaborate the last sentence please?

Is the REST channel connected to the docker image we would create or?

I understand every of these concepts independently but I don’t understand how they intertwine.

So basically each piece has rasa-supported ones, but can also be customizable, e.g. for your front end, you can connect to twilio, facebook, slack, etc., or you can create your own channel. To do this you would have to create a python module for your channel (usually you would build this off of the REST channel), and then add the channel to your endpoints.yml and make sure that file is also mounted by putting it in the $(pwd) directory that also gets mounted to the app folder.

A similar thing would be for your back end, which you would incorporate here, except you wouldn’t use the sdk image if you weren’t using python.

The way I would recommend to do this is to work out your custom input channel and/or action server locally, and then once it is all working add it to the docker-compose and go from there.

See also Custom Connectors


1 Like