Microsoft Bot Framework bot that uses Luis.ai to detect the user's intent and if the intent score is low or the intent "none" is detected, hands off to QnAMaker.ai
Log in to Luis.ai and create a custom model. Train and publish the LUIS.ai model and copy the URL presented to you by LUIS.ai on publishing. Note that {APP_ID}, {SUBSCRIPTION_KEY} and {BING_KEY} need to be replaced with the real values presented to you on the LUIS.ai publishing page:
https://{AZURE_REGION}.api.cognitive.microsoft.com/luis/v2.0/apps/{APP_ID}?subscription-key={SUBSCRIPTION_KEY}&spellCheck=true&bing-spell-check-subscription-key={BING_KEY}&verbose=true&timezoneOffset=0&q=
Copy it to the web.config file.
Log in to QnAMaker.ai and create a new QnA Service. On your "My QnA services" overview page (https://qnamaker.ai/Home/MyServices) get the Sample HTTP Request by clicking on Sample Code -> "View Code" beloning to the QnA service in question. find the {qnaKnowledgebaseId} and {qnaSubscriptionKey} and copy them to your project's web.config file.
POST /knowledgebases/{qnaKnowledgebaseId}/generateAnswer
Host: https://westus.api.cognitive.microsoft.com/qnamaker/v2.0
Ocp-Apim-Subscription-Key: {qnaSubscriptionKey}
Content-Type: application/json
{"question":"hi"}
Log in to the Azure Portal and register a new Bot Channels Registration. Copy down the Micrsooft App ID and Microsoft App Password and copy them to the web.config file of the main project.
<add key="MicrosoftAppId" value="" />
<add key="MicrosoftAppPassword" value="" />
Deploy the bot project as an Azure Web App and note it's endpoint in the following format:
https://[yourname].azurewebsites.net/api/messages
Copy this endpoint to the bot framework configuration page's "Messaging Endpoint" field.
Test your bot in the interactive chat window in the bot framework page.