Meet the Startup Using AI to Hang Up On Bad Customer Service
Chatbots are smart enough to handle 60-70% of routine service inquiries, so why are you waiting on hold?
For many consumers, the only thing worse than a confusing product is having to call customer service to figure it out. We all know the drill: long waits on hold, constant requests for personal information, inconsistent and uncoordinated responses – all to get the answer to a question that hundreds of other customers ask every day.
Given that kind of experience, it’s no surprise that only 12 percent of people would choose to contact a service provider by phone, according to a recent Internet Trends report. But despite the poor experience, and the fact that phone-based service costs companies between $6 and $20 per contact, more than 90 percent of service is still delivered by phone.
New technologies like artificial intelligence (AI) and voice recognition hold out the hope of a solution. The challenge is that AI systems need to be trained to understand human voices, parse out different kinds of questions, and formulate the correct responses. That training takes massive amounts of data and lots of time, even when the underlying technology is sophisticated enough to handle the task.
Wysdom.AI, a Toronto-based startup, wants to offer companies support for lousy customer support. They’ve not only built an AI system optimized to manage customer care across every channel – phone, chat, email, social and web – they’ve pre-trained it with three years of customer data drawn from real-world, demanding conditions.
With Wysdom comes experience.
Wysdom’s cofounders Ian Collins (CEO) and Jeff Brunet (President) have been developing solutions for customer support since the 00s, when they helped telecom providers manage the huge escalation in support needs that came with transition from ordinary handsets to smartphones. Their first business, Mobile Diagnostics, was successful enough that they sold it to HP.
Collins and Brunet launched their new company, originally called Crowdcare, to capitalize on the potential the emerging technology of natural language processing in support and service scenarios.
“Many of the contacts companies receive from customers are routine requests for the same information,” says Collins. “We realized a well-trained AI could take the burden of those contacts off call centers and phone representatives, giving customers the information they need through the channels they prefer.”
The team worked on the Wysdom platform two years before landing a critical engagement with Rogers, Canada’s largest telecom provider.
“Rather than just selling them the product, we operated it on their behalf and used it to train our AI,” says Brunet. He explains that Wysdom used the data from millions of customer engagements to identify different patterns and permutations in service requests, optimizing the system as it operated.
Knowledge is power.
That set of knowledge, which Wysdom calls the Corpus, forms the basis for the company’s offering and its advantage over competitors.
“There are a lot of companies offering bots that can engage with customers on social media in service scenarios,” says Collins. “Those might be good enough for simple use cases, but without data, they won’t be of much value, and might even frustrate customers they are meant to help.”
Collins says Wysdom is ready to handle a high volume of complex inquiries on day one, and customers will start seeing results in weeks versus the 12-18 month timeframe for customizing and training other solutions. He anticipates the system can skim off 60-70 percent of common inbound issues, with the potential of getting smarter, faster and better every day.
In search of lost time.
When you scale that up to enterprises that handle upwards of a million contacts per day at an average cost of $10 per contact, the savings are enormous. But the company insists that it’s not just a matter of financial efficiency.
“Customers waste billions of hours per year in frustrating conversations with call centers,” says Collins. “We want to give them that time back, by giving them good answers through the channels they prefer – Twitter, text, email, or whatever the next thing happens to be.”
Collins and Brunet recognize that this kind of technology might seem threatening to workers in the industry, but say it can actually help on that front as well. “Call center jobs are boring and repetitive when so much time is spent answering the same thing over and over, just reading from a script,” says Brunet.
“With our platform, you will still need people to handle support, but the calls that get through to them will be the interesting, challenging problems. It turns those positions into much higher value roles, and lets customer care professionals use their skills and talents on jobs that only a person can do.”
Press ‘one’ to continue.
Wysdom has been in the market for over three years, which is a very long time in such a new space. Over that period, they’ve raised $15 million in capital and signed up an impressive range of customers including Fido, Rogers, Virgin Mobile, Digicel, C Spire and many others, growing their team to 50 FTEs in four offices.
The recently rebranded Wysdom platform is cloud-based but can be installed on premises if customers prefer. Collins says their product supports English, French and Spanish, with additional languages on the way.
In August, 2017, they were invited to participate in the Microsoft Accelerator in Seattle to further hone their technology and business strategy.
“There aren’t many areas of business that need disruption more than customer care,” says Collins. “Customers are saying loud and clear that they want channels beyond the phone to get their questions answered. We’re here to help companies say ‘yes’ to those demands.”