Sep 12, 2018

Chatbots and the Future of Customer Service

P.V. Kannan
P.V. Kannan

Co-founder and CEO

Chatbots and the Future of Customer Service

I began 2018 with a few predictions about the ways I expected AI to shake out this year. One of the trends I discussed was companies embracing the concept that humans would work hand-in-hand with artificial intelligence (AI) and chatbots to improve the customer experience. I am happy to say that’s beginning to happen.

Digital transformation remains a top priority for businesses, and consumer-facing companies are deploying AI chatbots to automate customer service. They are doing so by deciphering huge amounts of customer data, comparing on-domain and off-domain data, and running that data through algorithms to determine consumer intent. Once they do, they can personalize the customer experience. Only by using artificial intelligence can companies process this immense amount of data. It’s simply not possible to apply enough humans to do this kind of analysis.

But companies can’t live by AI alone. By itself, it cannot succeed. A simulation of human intelligence by machines, designed to perform human like tasks, ironically requires a lot of support from humans. A critical criterion for digital self-service is to maintain the quality of the interaction. Failed digital self-service leads consumers down an unhappy path, and ultimately to abandonment of self-service channels. For the business, this results in increased call volumes, which defeats the purpose of automation.

Human Guidance Is Critical

AI is far from taking over the world. Rather, it is like a child that needs constant education and supervision to learn and perform this task of training. For the foreseeable future, humans will have to feed AI systems with experiences and situations to ensure the machine has references and large resources of knowledge to draw from. Humans will have to regularly intervene to provide the answers that chatbots can’t, or don’t feel confident about. In most cases, chatbots are not currently capable of handling and completing complex transactions that ensure a customer’s overall experience is both smooth and satisfactory.

The most important aspect of the human intervention, however, is empathy. Human agents are the ones who differentiate a brand’s customer service, not the chatbots. While some chatbots are being trained in sentiment and emotions, it only helps them deal with rudimentary problems, while more nuanced interactions will require a human. I firmly believe that chatbots will get there, but this will require extensive training by humans, using real human experiences, over a longer period of time.  

By itself, AI cannot succeed. A simulation of human intelligence by machines, designed to perform human like tasks, ironically requires a lot of support from humans.

Chatbots and Humans Both Benefit

One of the biggest beneficiaries of blending bots and agents will be your agents themselves. Not only will chatbots augment agent tasks by providing data, insights and analysis; they will reduce the average handle time (AHT) to ensure more interactions are completed with more efficiency per interaction. But in the digital age, AHT really isn’t the best metric. It’s about Total Customer Interaction Time. This decreases, too. The bot-assisted augmented customer experience ensures less waiting time for consumers, thereby increasing overall self-service adoption, which is critical to digital transformation.

Once chatbots start handling the more mundane and repetitive customer support tasks, human agents can focus on developing their skill sets and concentrate on high-value tasks which will lead to increased overall agent satisfaction and positively impact customer retention.


An example of blended AI would be a shopper trying to buy a product. After being guided by an informational bot, let’s say the consumer has to turn to a live agent for help when the credit card is charged for an incorrect transaction. Here’s an example of how that might play out:

Chelsea is looking to get into photography and googled DSLRs for beginners

After browsing the list of results, review sites, and blogs, she finally ends up on Shop Electronics.

Chelsea bought her phone at Shop Electronics a month ago and is automatically logged in to their website

VA: Hello Chelsea, I’m the Shop Electronics Virtual Assistant. Are you contacting us about your phone or something else?

Chelsea: I’m looking to get into digital photography.

VA: We have a great selection of entry-level DSLRs we can show you, would you like to view our top-rated options? [View Options]

Chelsea clicks to view options, reviews, and adds to cart.

VA: Great choice! We offer an extended warranty; would you like to add it to your order?

Purchase is completed.

Later, Chelsea sees her credit card bill and notices a charge that she did not approve

Chelsea opens Shop Electronics VA on her mobile phone.

Chelsea: You charged me for something I didn’t want.

VA: Sorry! Let me look into this right now

Image in corner changes from a bot to a human.

Chat Agent: Hello Chelsea, I understand you have been charged for an item you didn’t want – there are two purchases made in the last month that are both pending delivery. I’ve pushed a list to your screen so you can click on the purchase that you didn’t want.

Chelsea clicks on the second item.

Chelsea: Why did this happen?

Chat Agent: You may have added this item to your cart by accident and purchased it when you used our “Quick Checkout.”

Chat Agent: Would you like me to adjust your settings to have the system double-check your order before final purchase?

Chelsea: Yes, please!

How do chatbots learn from humans?

In this new world, chatbots learn from human agents as well. Every time the chatbot transfers to a human, the human tags the interaction so that the bot can learn. Not only does the chatbot learn to better interpret the consumer’s questions and comments, but it also learns to take a more educated guess at what the right response might be. Over time, the bot learns to emulate the skills of a company’s best human agent. This turns agents into “super agents,” supported by a great team of bots behind them.

Chat details

The right blend of virtual agents and human chat agents can not only lead customers to a happier path, but can also yield significant cost savings, and even increase sales. This is just the beginning of a collaborative and symbiotic customer experience ecosystem that will witness humans and AI working side-by-side.

Contact us if you’d like to know how agent and bot interactions could be applied to your company.

Related Posts

Ethical Issues/Factors in Chatbot Personality Design
Jan 12, 2021 Too Human for Humans: Ethical Issues in Chatbot Personality Design

Let’s dive into some of the…

How to Design Your Conversational AI Chatbots
Jan 8, 2021 How to Design Your Conversational AI Chatbots

Armed with the latest natural…