Zendesk Launches Automatic Answers for Contact Centers



Zendesk has launched Automatic Answers, a feature powered by machine learning to help customers solve their inquiries without having to go through contact center agents. Companies can embed the technology right into their Web sites or mobile apps.

Automatic Answers woks before customers submit their inquiries to contact center agents. It scans the content of the query, natively auto-responds to customer tickets, and uses machine learning to identify articles within the company knowledge base that could answer the question.

"It auto-replies with a list of articles that could answer the question," says Sam Boonin, vice president of analytics at Zendesk. "And with machine learning, answers get better over time based on an analysis of which answers have worked and which ones did not."

If a customer indicates that his inquiry has been solved, the ticket is closed. Tickets that remain unsolved proceed to the customer service team as normal.

Automatic Answers is currently set up to work with email inquiries, but Boonin says the goal is to open it up to other channels. "It's a technology that we'd like to use across all channels as it gets more mature," he says. "Email, chat, social, SMS—we want to be able to apply it across them all."

The solution works on PCs, tablets, and mobile devices and supports all standard email clients.

According to Boonin, Automatic Answers is Zendesk's latest foray into machine learning." The company, he says, has been working on machine learning for more than a year and "is unique in the way that it is applying the technology to customer inquiries."

"We're excited by the fact that so much buzz is happening around machine learning, AI, and bots. We want to use the technology differently and apply it to very specific contact center use cases," he adds.

The goal of Automatic Answers is to deflect basic inquiries away from customer service agents, something Boonin says "is easy to do when you control the interaction this way."

"Zendesk continues to innovate its machine learning capabilities to help businesses provide effortless customer engagement," said Adrian McDermott, senior vice president of product development at Zendesk, in a statement. "Automatic Answers' predictive capabilities provide customers with the resources they need to solve their issue quickly, and helps businesses free up their agents to focus on inquiries that need a human touch."

Zendesk made Automatic Answers available to several of its customers through an early-access program. One of the companies to test the product was Plex, which provides a suite of products and services for accessing media across devices.

"At Plex, we're always looking at how we can use technology to smartly serve our customers and improve the overall experience," said Scott Olechowski, chief product officer at Plex, in a statement. "We're excited about Automatic Answers because we hope it will give our customers the answers they need to solve their questions faster."

Another stated goal is to make the answers better and more relevant.

With normal auto-responders, that is the case about 50 percent of the time, according to Boonin. "With machine learning and AI, we hope to get that up to 80 percent or 90 percent," he says.

A lofty goal, but Boonin thinks it's possible. "With AI and machine learning, we'll always be working to make the answers better," he says.


Related Articles

Satisfaction Prediction uses machine learning to identify which customer interactions are at risk of turning bad in real time.

Posted October 07, 2015

With Zendesk SMS, companies can send and receive text messages through Zendesk.

Posted May 13, 2016

Online review platform provider Trustpilot is now integrated with Zendesk, a cloud-based customer service platform provider, in an effort to empower Trustpilot clients with a wider array of customer service tools and technology. Through the integration, a review on Trustpilot automatically generates a customer service ticket in Zendesk so that agents using Zendesk can be immediately alerted and can appropriately respond to the review.

Posted September 16, 2016