Build And Publish A Bot Using QnA Maker


In this article, I will explain about creating a bot that can answer the questions we all ask. We will achieve this by creating the question and answer template, and the QnA maker website. As there are many types of bots like LUIS etc., one of the types is FAQ bot. This type of bot is not well documented and not advertised very much. However, this is very useful in conversational knowledge to use this type of bot. We can create our own QnA/knowledge/FAQ base bots, and another happy news is that this doesn’t require any coding. The QnA Maker doesn’t require any bot framework. It is a template in bot service that enables bot's rapid intelligent and development.

Steps for using QnA Maker for creating a bot

Step 1

Log into

Step 2

Click "Create new service", and accept the terms and conditions by selecting "Continue".

Step 3

Now, we need to provide a name for the service and the URL for FAQ. For example - I have provided a website name as Microsoft that provides the customer service like

Step 4

Create a FAQ file in formats like PDF, .doc with the file size limit of 2 MB. The file with FAQ will be imported. In this example, I have uploaded a text file with questions and answers. The article may vary from the uploaded FAQ files.

Step 5

Click "Create" for creating a knowledge base for the service.

So far, we have created a bot. Now, we have to train the bot and then we will publish it.

Step 6

Click "Test" in the left side of knowledge base. Now, we can see the content of the database on the testing window.

Step 7

In the test dialog box, the chat with the bot can be done and the bot responds with appropriate answer from the FAQ database. The bot will respond to the question after getting trained with answers to particular questions.

Step 8

Click "Save and retain" after the completion of training.

Step 9

Now, after the training process is completed, I would like to publish the bot. For that,  click "Publish".

Step 10

Click "Publish" again and the bot will be published.

Thanks for reading.