Azure Artificial Intelligence (AI) Series: Building a Chatbot

This is the first post in our series on Azure Artificial Intelligence

Chatbots 101 – Quick and Easy Customer Service Chatbots

This is the “very” first article on Azure & AI series.

Audience: Software developers with some background of Azure and Microsoft Products. Have little or no idea how to build a Customer Service Chatbot.

What is it about?

Overview on how Power Virtual Agent can be leveraged to create Customer Service Bots that handles Dynamic 365 Ticket lifecycle.

What it is not about?

Here, we are building a minimum viable product – no custom AI. If interested in intelligent Bots, stay tuned for upcoming articles in the series.

Why we wrote this?

We want to introduce you to the capabilities provided by Microsoft Power Virtual Agent, that you may be interested to leverage.

We want you to get a familiar and comfortable with the Microsoft Power Virtual Agent.

Creating a Chatbot

Let’s get started.

Pre-requisite: Access to the Power Virtual Agent. An up and running Dynamics 365 environment.

Go to https://powerva.microsoft.com/and login using your organizational credentials.

On the top right, click the Bot icon  

and then click ‘New bot’

Give the Bot a name, choose language of Bot’s communication, and select the environment from the dropdown, where your organization stores, manages, and shares the business data, apps, and the flows. Hit Create.

Azure Artificial Intelligence

Create a Topic

Create a Topic. A Topic gets triggered once the chatbot customer enters a Triggering phase.

Once the Bot is created, hit Topics and then click ‘New Topic’ button  

Provide Name of the Bot, and the Triggering phrase which will start the Topic. Save the topic.

Azure Artificial Intelligence

Click Go to authoring canvas to define step by step process of how this Topic will work.

Now click the + sign under triggering phrase node to add the next step (node).  In general, your options would be few or all of Ask a question, Add a condition, Call an action and Show a message.

Azure Artificial Intelligence
Azure Artificial Intelligence

Let’s discuss a few nodes implemented here.

Create a question node to choose among the sub-topics as below, and store response in a variable.

Azure Artificial Intelligence

Then create branch-based conditions and the follow up nodes to the Add to RDS App Group condition node as below:

Azure Artificial Intelligence
Azure Artificial Intelligence
Azure Artificial Intelligence

When creating an action node, you can tie it to a flow.

Azure Artificial Intelligence

In the above example, the flow name is RDS. Let’s see the mechanics of the flow. Flow provides a number of trigger and actions, with the glimpse as follows:

Azure Artificial Intelligence

In this specific example, the flow takes input from the chatbot, then create a ticket in Dynamics 365, and return ticket number to the user to which chatbot is communicating. Initiate an approval request (email) to the approver. Once approved via email, calls an Azure function is called via HTTP to perform actions related to the ticket.

Azure Artificial Intelligence

Details of some of the actions and triggers are as follows:

Azure Artificial Intelligence
Azure Artificial Intelligence
Azure Artificial Intelligence
Azure Artificial Intelligence

Test your Chatbot

Once the flow is set up, save it and then test the Bot. Sample results are as follows:

Azure Artificial Intelligence
Azure Artificial Intelligence
Azure Artificial Intelligence
Azure Artificial Intelligence
Azure Artificial Intelligence

You now have a customer service Bot up. These Bots can be published to multiple platforms including Websites and Microsoft Teams. For more information, see the references.

This concludes today’s topic. This is the first post in Azure Artificial Intelligence series. Please stay tune for more posts.

If you need more information or looking for a customer service Bot solution on Azure/Office 365, contact Wintellisys!

References:

Publish your chatbot. https://docs.microsoft.com/en-us/power-virtual-agents/publication-fundamentals-publish-channels

Learn Dynamics 365. https://docs.microsoft.com/en-us/learn/dynamics365/

About the Author Samia Sherwani

Samia Sherwani is the Director, DevOps/Security at Wintellisys. Her areas of expertise include Enterprise Architecture and leadership, Data Science, Big Data Administration and Development, AWS/Azure/Blockchain Architecture & Development, and DevOps. MS Computer Science, MS Project Management, & MBA. Certified in AWS & Azure Architecture, AWS Machine Learning, Azure Security, Hadoop, Java, Data Science, and Blockchain.