Forbes Contributor, Murray Newlands interview with Agent.ai CEO, Fred Hsu
Featured on Forbes, August 28, 2017
Stories of human-level artificial intelligence have been fashionable as our technology increases, with AI as the focal point of countless dystopian future novels and movies. The fear that artificial intelligence takes over drives our collective unconscious. But as we move further into the digital age, the evil AI trope has been left behind, in favor of the truth: benign workhorse chatbots and digital assistants are changing customer service as we know it, for the better.
Since the future is already here, and helpful chatbots are easy and cheap to adopt, the question remains: why do so many apps provide such poor customer support? Many brands insist on using slow, frustrating email contact for customer questions and issues, and consumers are increasingly turning their back on those brands in favor of instantaneous, painless problem solving using messenger bots.
Fred Hsu, CEO at Agent.ai, recognized that the consumers were turned off by this frustrating customer service snafu, and set out to rectify it. Hsu’s company started to address not only customer pain, but customer service agent pain. What software could increase productivity, allow for critical customer info and history to be accessed immediately, and sort customers by lifetime value so their inquiries would be addressed first? Agent.ai was the answer, and Hsu was kind enough to tell me all about it in a recent sit down.
How did you start the company? How did the concept come about?
The concept for Agent.ai started out of my own user pain. I was spending about $600 a week on a very popular rideshare service, and I was also spending about $5000 a year on very popular games like Clash of Clans. I noticed a trend where high-value users like me, or whales, were getting bad service. Whenever I needed service or had complaints, like let’s say in Clash of Clans I lost my village just because I upgraded my phone from an iPhone 6 to a 6S, or my driver picked me up late, I noticed this really clunky support process. Those apps, as well as other apps, would send me to an email submission form, and maybe I would get a response in three to five days. It was not in alignment with how I felt I should be valued as a user.
I also noticed when I was in the mobile marketing business with my prior company, that we would be tasked with sometimes charging customers $20 to $30 per high-value user within a game. I would do that marketing hand off and provide that customer for these large advertisers, and I would be scratching my head, because they would be spending all this money acquiring a high-value user, and then giving them what I considered to be suboptimal customer service. I wanted to fix that gap from both angles.
How did it develop? What was the first stage of putting it together?
The first stage was to visualize the experience from the end user. End users, in our experience, want great service, instantaneously. They want it in a timely fashion. People are becoming more messaging focused, so what they used to want in terms of service on phone or email, these days they expect to get over messaging.
It was first about figuring out what the end user would want, certainly using myself and my team as examples with their own internal pain. Then I turned to Barry Coleman, my CTO, who has an extensive chat CRM background over ten years with a company called ATG/Oracle. He is also the AI brainchild behind our company, helping us achieve the goal of fixing the customer’s experience and then getting the right team in place.
So the big question, how do you think AI will change or has changed customer service forever?
First, you’ll never have to repeat yourself again; better information gathering, in context and with full sharing, means you only have to provide your information once. This makes customer support agents will be far more effective. The AI will support human agents by automatically handling simple informational lookups, and by putting the right customer information at an agent’s fingertips, equips them with the data they need and can even recommend a course of action. Common, yet complex inquiries will be handled faster, around the clock. API lookups can be made more quickly by machines than by human agents, and can be made at anytime.
Increasingly, support will get friendlier and more stable: since AI can take care of so many support questions simultaneously, standards will move towards support quality over speed or deflection. AI can help enforce these standards by gauging customer happiness levels. AI will free support agents from having to handle mundane, repetitive, simple inquiries, allowing them to focus on solving more complex customer problems, reducing service team churn.
Training AI to handle support inquiries will be near instantaneous, as AI can ingest and leverage FAQs, product information, and past support tickets immediately compared to human agents who need days or weeks to review and learn the same material. Agents will be able to more quickly step into new support roles, servicing and offering new products with AI suggested responses. Like a great employee, the AI learns and improves over time, and agents and AI will learn from each other.
Also, AI is channel agnostic, which means you can choose your channel: chat, email, phone, or social. You can use the communication channel you most prefer, and support will be available 24/7/365 — even for the smallest companies.
What do you see as the optimum positioning for AI in the customer service space?
What I see is, on one end of the spectrum, there is what I call dumb chatbots, and on the other end of the spectrum, there is human-intelligence-level AI. I think people are fearsome/curious about where the future lies with AI. Are the machines going to kill us? Those are the type of early-dramatized descriptions that accompany the fear of the unknown AI future.
But in the middle, Agent.ai sits in this bucket called “AI assists.” In our view, what’s different about us from the many companies out there is that we focus in on good human agents, who have the right AI tools. We don’t want agents that have to look up ten different systems. With AI assists, they can handle ten to 20 simultaneous chats or more, rather than two or three, which is about the industry standard. In short, we’re all about AI assists for the human agent.