Build an assistant to evaluate various health insurance policies


This code pattern demonstrates a way to build an assistant that can answer queries regarding COVID-19 insurance policies of various companies. It also compares the policies and add-on-related queries given by them. The application is developed using IBM Watson® Assistant and IBM Watson Discovery.


Various insurance providers have created insurance policies specifically for COVID-19. With this, it’s very hard for people to select an appropriate COVID-19 insurance policy for themselves and their families. The application created in this code pattern guides you through various COVID-19 insurance policies so that you can explore them and make decisions for yourself and your family. The application also compares the policies and add-on-related queries given by them. In Watson Assistant, we create Dialog skills, then use the Search skill capability to develop a model in Watson Discovery that will understand the policies and provide an interactive interface. The application is beneficial for various insurance brokers and also to people who want to know and compare the policies of various companies.


Build an assistant to evaluate health insurance flow diagram

  1. Upload documents to Watson Discovery to use Smart Document Understanding to train the model on various policy documents.
  2. Watson Assistant interacts with the user to gather personal details.
  3. After collecting personal information, the Search skill is initiated.
  4. The results are fetched from Watson Discovery, where the model has been trained on various policy documents.
  5. The results are returned to the user.


Find the detailed steps for this pattern in the readme file. The steps show you how to:

  1. Create a Watson Assistant instance and create a Dialog skill.
  2. Set up the Watson Discovery service.
  3. Create a new collection.
  4. Upload the documents.
  5. Integrate the Search skill in Watson Assistant.
  6. Copy the Integration ID and update the ID to Flask.
  7. Run the Flask app.



Related post