Wise.io launches automated support replies for Salesforce and Zendesk

Wise.io, a startup that brings machine learning smarts to cloud software, is announcing today the launch of a new feature called Auto Response. The idea is to make customer support teams more efficient — by generating and sending automated replies to customers’ inquiries.

Auto Response integrates with service desk applications like Zendesk and Salesforce’s Service Cloud. After training on customer support tickets and agents’ replies to them, the system can come up with what it thinks will be good responses to new tickets. If the system isn’t very confident that it knows what answer to give, it will offer recommended responses for agents to build on.

The feature calls to mind Smart Reply, an interesting piece of the Inbox by Gmail email client Google introduced in November, which offers up quick replies for people to select while on the go.

Jeff Erhardt, Wise.io’s CEO, isn’t particularly interested in using Smart Reply for his personal email because he likes to make sure his “voice” comes through. And in the case of his personal email, the volume is low enough that responding doesn’t take up too much time. But he recognizes that a feature like Smart Reply could certainly play a positive role for some people.

“If it was a requirement for me to reply to every single email that I receive, I would absolutely love and think very much about Auto Reply,” Erhardt told VentureBeat in an interview.

Suddenly the idea of a simple response makes a whole lot of sense. No wonder service desk software company Freshdesk bought Frilp last year. Automation could very well come in handy when it comes to answering many people’s questions.

Beyond customer service, machine learning is being incorporated into other types of enterprise software, too. Following its acquisition of Upshot, human capital management software company Workday is focusing on ways to predict retention risk by looking at what people say about their jobs — something that isn’t always easy for companies to do well when they’re looking at employees one by one. And in the area of customer relationship management, Salesforce acquired RelateIQ and has been integrating the startup’s technology in order to automate the process of data entry.

This isn’t the first feature that Wise.io has introduced to bring machine learning to existing software. It can already recommend template responses for tickets and route customer service inquiries to the right agents.

Auto Response gets smarter as time goes by. That is, within a month, or even a few days, the system will detect emerging types of tickets and the responses that agents give to them. It all comes down to the number of tickets coming in, because that’s an index for the frequency of certain patterns in interactions.

But it takes just a day or two for Auto Response to start working. The system trains on historical tickets and then poof, it’s ready to answer questions, said Erhardt, formerly chief operating officer of Revolution Analytics, a company commercializing the R programming language, which was acquired by Microsoft last year.

The feature is getting rolled out to all Wise.io customers. The cost is determined based on — you guessed it — the number of tickets a customer receives each month.

Wise.io started in 2012 and is based in Berkeley, California. The startup announced a $2.5 million funding round in March 2014. It has several dozen customers, including Invaluable Auctions, Republic Wireless, and Thumbtack.

One customer that has been using Auto Response leading up to today’s release, Product Madness, has found it to be useful so far.

“In addition to auto-responding to low-level tickets, Wise.io helped us improve response times across the board because its software also suggests answers to tickets based on their predictive models,” Product Madness vice president of marketing Francesca Noli wrote in an email. “The agent simply edits the best answer and replies, therefore cutting response times. This feature has also positively impacted training for new agents.”