This starter pack is to help get started on insurance related tasks. The starter pack demonstrates a few different areas where a Rasa assistant can help with insurance-related functions, getting information for an insurance quote, handling common tasks like ordering a new ID card, and helping customers manage their claims.
Before training your bot make sure you have installed all of the requirements from the requirements.txt file:
pip3 install -r requirements.txtTo run the bot locally you must first train your bot by opening a terminal window in the project directory. In the terminal window enter:
rasa trainThis will train a model for your bot that you'll be able to interact with in the next steps. You can find out more about rasa train in the Rasa docs!
When you're training successfully completes you will have a model artifact. Before you can start interacting with your bot the Action Server also needs to be running. The action server handles custom processing of messages. Starting the action server requires opening a new terminal window in your project directory. In the window enter:
rasa run actionsYou will see a listing of the different actions that are a part of the server. You will need to keep this terminal window open.
Finally, the last piece is to start a Duckling server. The Duckling server will help the bot robustly extract numbers from the user messages. Open one more terminal window in your project root and enter:
docker run -p 8000:8000 rasa/ducklingSimilar to the Action Server keep this running while you interact with your bot.
Now you can talk with the bot! In a terminal window enter:
rasa shellThis command will allow you to talk with the bot. If you want more more detail about what's happening with your bot you
can add --debug to the command to display all of the debugging information.
Right now the bot accomplishes these core insurance functions:
- Get a Quote
- Order a New ID Card
- Check Claim Status
- Pay Outstanding Claim Balance
- File a New Claim
The demo currently has mock data for a customer with a handful of claims to scroll through to demonstrate
The bot is ready to be deployed to a Rasa X instance. The easiest way for you to deploy your bot to Rasa X is to utilize the one line deployment script.
Once Rasa X is running you can use the git integration to load your bot into the Rasa X instance.
You will need to have docker installed in order to build the action server image. If you haven't made any changes to the action code, you can also use the public image on Dockerhub instead of building it yourself.
It is recommended to use an automated CI/CD process to keep your action server up to date in a production environment.