Docker Compose configuration for rapidly getting started with running a bot using the Rasa Pro Developer Edition. This is intended as a lightweight way to deploy a Rasa assistant that will receive only small amounts of traffic.
Note you will need an extended license to run Rasa Pro Developer Edition on a server.
-
Create a VM on your cloud provider of choice, and install Docker. 1vCPU and 2GB RAM should be sufficient unless you've trained a big NLU model.
-
Create a directory and copy in the following files from this repo:
docker-compose.yml
endpoints.yml
credentials.yml
actions.py
models/model.tar.gz
- Create a file called
.env
in the same directory with the following contents:
RASA_PRO_LICENSE=<your Rasa Pro license goes here>
OPENAI_API_KEY=<your OpenAI API key goes here>
-
Deploy the bot by running
sudo docker compose up
. (On older docker versions, it would besudo docker-compose up
) -
Send your bot a message by sending a POST request:
curl -XPOST http://<host>:5005/webhooks/rest/webhook -d '{"sender": "test", "message": "Hi"}'
Now that you have a demo assistant deployed, here are the next steps you might want to take:
- Upload your own model to replace the default one in the repo.
- Add your own custom action files, or build a custom action server image
- Configure your assistant to respond on other channels by
modifying the
credentials.yml
file, see here