My thoughts on Virtual Agent for Customer Service

When Microsoft released the Bot Framework 3 years ago, the goal was to ultimately empower any developer to build and deploy the bot to any channels (FB Messenger, Teams, Skype to name a few) with ease to revolutionize the way the customer service operates to realize efficiencies and drive down the operating cost.

The Bot Framework provided the developers with the wholistic platform which unsurprisingly came with the learning curve, that was smoothened by the number of samples and amazing documentations and a wonderful community around it.

But here are the challenges:

  1. We need a pro developer to build the bots.
  2. Building bots require us to renounce our traditional web-app mindset and learn best practices that it demands.
  3. Time to the market is relatively longer.
  4. Learning additional services such as QnAMaker, LUIS and Dispatch to build meaningful bots.

So, did it revolutionize the customer service as it was originally intended? – While it has the potential to revolutionize, building bots was a bit overwhelming for many, unless you are developer.

Can something be considered democratized if only a subsection of the community has access and familiarity towards it? If the answer is “No”, how can we democratize the technology that is very sophisticated and complex in nature? Is it even possible?

Let’s take a step back and talk about Microsoft Flow. There were a number of new technologies introduced and/or made popular in the recent days by Microsoft, out of which Flow is by far the best product that is so powerful which can turn the very complex integrations into something that is very trivial. The most important point to highlight is that even the non-developers can easily build flows within minutes that the pro-developers traditionally take a month to build. Community loves Microsoft Flow because of the power it yields and the ease of use.

Microsoft takes the Flow that the non-developer community loves and combines it with the Bot Framework that is already very powerful. In this marriage, Virtual Agent for Customer Service is Born.

Dynamics 365 Virtual Agent for Customer Service empowers customer service teams to easily create powerful bots using a guided, no-code graphical interface without the need for data scientists or developers.

Bot Framework provides the base platform, whereas Flow acts as a means of supplying intelligence to the virtual agent via services integration. In my opinion, Virtual Agent for Customer Service is by far the best application of Microsoft flow to date.

If you know how to Flow, you know how to Bot.

This is the 1st iteration of this product, so it is not perfect. Authentication has to be sorted out, branding the bot is not available right now. But the functionalities that it currently offers can be leveraged to create and deploy a bot to production that is whole and immensely useful.

The Bots are the evolution of the consumer interaction. When paper forms evolved to web forms, the business had to leverage the behaviours that is innate to the web platform in the form of best practices such as business rules/validations, process optimizations, analytics and user trends. The evolution to the web platform also brought in challenges such as discoverability, rigid process and some process cannot be digitized on the web platform for various other reasons.

This led to the evolution of the Conversational UI to solve some of the problems that were previously unsolved by the web app platform. Conversation UI platform brings in a lot of benefits like its predecessor such as high discoverability, shorter engagement time, natural conversational ability. But to reap these benefits, one should let go of the web forms developers’ mindset, but rather think from the perspective of the bot platform. This means, we need to throw away some of the practices that we have been following as a web developer, which is obviously a huge challenge and any help that the framework can provide will be highly appreciated. As this is a huge topic in itself, I have compiled and blogged about the best practices. This is where the virtual agent for customer service shines and/or will shine in the future. As the product is tightly coupled with the other services and fully controlled by Microsoft, the bot builders do not have a way to build a bot that is sluggish in nature.

Let’s take an example: I consider “greeting the user” to be a best practice when the bot initially engages with the customer. This also provides an opportunity for the bot to summarize its capabilities to set the tone of the conversation. Virtual Agent for Customer Service has an in-built Greeting mechanism that one can customize per their needs – Problem Solved.

I also think if the bot does not know how to handle a particular customer’s query, it needs a way to route the customer to the real human agent. The Virtual Agent for Customer Service has the extensions for hand-off out of the box.

Suggested options are one of the important practices in building the bots. In the below example, when the customer was looking for the Store Hours, the bot provides clickable button with the store names, guiding the user to keep the conversation focused and narrow. This speeds up the engagement as the user has everything they need right upfront to continue the conversation.

The bot builders can organize various functionalities that the bot perform in something called Topic which is modular is nature and easier to maintain, which is great.

I can finally say that the gaps and challenges that we once had in order to democratize the Bots are now being rapidly closed and I can visualize a number of ways this can disrupt the way the businesses work with their customers. The dream of AI for every business is not impossible anymore, if we are up for it.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s