In this step we augment our conversational AI applications capabilities by integrating Azure AI Search to create a composite AI solution.

The main processes involved in this step include 5 main parts:

• Part 1 - Creating the filter query to send to the Azure AI Search Instance
• Part 2 - Constructing the HTTP API request to consume the Azure AI Search service
• Part 3 - Extracting facets Azure AI Search API JSON response
• Part 4 - Identifying the the most suitable answer
• Part 5 - Confirming if the user is satisfied with the answer
• Part 6 - Does user want to search for something else?

Part 1 - Creating the filter query to send to the Azure AI Search Instance

In Part 1, we define a property to store the facet filter. The first time the dialog is run this will contain a null value. This is because the value is set in a later part of the logic which restarts the dialog and is responsible for defining the value.

1. Click on the + node and select a Set a Property activity. Set the following values:

   • Property: turn.userQueryFilter
   • Value: =coalesce(dialog.userQueryFilter, "")

The designer canvas in UI Builder should now resemble the following: