Why and how an AI agent won a customer service award

Hi, and welcome to the CXPod. I'm your host, Liz Golgowski of the Customer Strategist Journal. Let's start this episode with a question. When I say dentist, what comes to mind?

Maybe it's that you need to schedule your next cleaning, or you're thinking of a memorable root canal, or maybe having braces as a kid like I did. But whatever you're thinking of, chances are it's not AI. Healthcare, medical, dental, is very personal, and the use of AI within the space is especially sensitive, which is why I'm excited today to talk to Nathan Christensen of Henry Shine, a leading provider of dental supplies and technology, to practices around the world. He's got some unique perspectives on how to embed AI, specifically customer-facing AI agents, in an industry where people and physical touch really matter.

So thanks for joining us today, Nate. Welcome to the program. Thanks, Liz. Excited to be here.

So before we dive in, can you tell us a little bit about Henry Schein, Henry Schein One, and then your role? Yeah. So Henry Schein One, where I work, is we're the largest dental software company in the world. So a lot of times people go to dental school, but they don't necessarily go there to run a business and have to run all the operations of it and so we're the one-stop shop or the microsoft equivalent for a dental office that allows the dentists to worry about their patients while we help take care of filing their claims helping with their insurance verifications helping with their ledger helping schedule their appointments and giving them kind of the full suite to solve their issues.

Great. So just to set a baseline, I do this with all my guests. How would you define what a good customer experience might look like for your clients and customers? For sure.

So our customers, a lot of them are older or more seasoned, experienced dentists, and they've been trained over the last 10, 15 years to call us just about for any kind of need or issue. And we've given them that white glove, hands-on kind of experience. And really our goal is to keep the human interaction in place. So we'll talk a little bit about AI and how we still do that with AI.

But we really haven't lost that vision of providing customers the issues they need resolved as fast as possible. So when my wife has me call the hotel or the airline, she hates to have her call that. So she gives me to call that. And really, at the end of the day, I want to get my issue fixed with the airline or with the hotel as fast and as quickly as possible.

That's the first thing that matters for me, as well as making sure it's accurate and done properly. And so the biggest thing is experience that we're giving our customers with high quality, like trust that's accurate, that's quick, that's efficient and resolves their issues. So how does then that play out as you start to incorporate things like AI within your customer service organization? I hear you've got the AI agents alongside human agents.

You haven't completely replaced them. So tell me about how you've approached using AI in that customer facing role. And, you know, this definitely could be kind of a prickly situation. Like you said, some of your customers calling in may not be used to it or potentially averse to it.

I'm just curious how you've kind of gone about using AI. Yeah, for sure. So I think first we can talk about what a not ideal customer experience looks like. And this was a couple of years ago.

We had an outage of one of the companies that we work with or worked with. It was a major health care outage that affected a lot of health care companies And as a result our phone lines got completely piled up in full Customers were waiting on the phone for 20 minutes And that not an issue you can resolve immediately because it takes months for us to train and onboard new live agents that can handle our 60 products that handle about 70,000 customers. So we had to come up with a scalable solution.

Now, a lot of people say, well, now implement a chatbot. And we do have chatbots that can help and answer calls and questions. But in the healthcare industry specifically, and where a lot of our dentists are experienced with answering and wanting to talk on the phone, we had to meet the customers where they were. And so we were able to explore multiple different AI, voice AI tools and different vendors.

the technology for some was not quite there and for others it was fantastic and we were able to choose one of the best ones that we could find and essentially we found that we could give this you know 15,000 knowledge base articles and overnight it would know everything about our product which helped answer a lot of the questions and alleviate a lot of that initial training burden and kind of the Q&A type of side of these calls. So today we're taking about 1,500 calls a day by an AI agent.

It's actually handling those calls in about a fifth the time as a live agent is. And then we still have those live agents on the other side that are redirected to by the AI agent, but now the customer is not having to restart. So this scenario would be is, hey, I have this problem with this product. Can you help me?

It will gather all the initial information. And if it can't resolve it for the customer, it now passes that information straight to the live agent. So is that a voice AI agent? This is a voice AI agent, yes.

So this is just a topic I'm curious about as I hear people talking about using some of these voice AI agents. Was there some decisions made in how you might personify that agent? Like, do they have a specific voice? Do they have a specific personality or using of humor or anything like that?

Did your discussion not only how to use it and use knowledge base, but also make it feel like an interaction a customer would want to have? Did that go into it? And if so, what kind of decisions did you make? Yeah, absolutely.

It was huge. And I would say as we first started, a lot of times our customers sounded more robotic than the AI agent did, which was interesting. And one of the big problems we had to handle and face is we are used to calling and yelling at Alexa or yelling at Siri to play our right song on Spotify. And we all have that experience where it plays the wrong song or we get on a phone tree and we're like, this thing is stupid.

Like, I'm just so frustrated. Get me to a live agent. And so initially our customers are accustomed to anything that's a fake voice is bad. Get me to a live agent.

And so we really had to work on the marketing, the personification and that intro message to gain their trust. So our agent, her name is Claire, and we use that because she provides clarity for our products. and then we've marketed to our customer hey we can now answer your questions 24 7 call us at midnight if you need claire will be able to answer anything and help you she has all of our products on her fingertips and knows things well and then we've really trained her voice as well she she's emotionally intelligent so one of the things that's interesting and this was a big thing as we were looking through a lot of vendors because there's a lot of fakers as well that claim to have a good voice ai solution and they're just building it on the side right now um and so one of the things we had to do is we actually had our support agents go and yell and curse and swear at claire and like put her through a whole ringer and if it was you or me and someone was doing that to me I probably would crumble or get mad or like how could you talk to me like this?

And Claire handles it like a champ. And so we really focus primarily on that emotional intelligence. The difficulty isn't getting her to have the knowledge of our product. It's getting her to gain the trust of our customers.

And you refer to the AI agent as her, right? You try to make it feel like another person in the mix. Another person in the mix, yes. Right.

Which then, I want to hear about, I had heard that you actually had Claire win a customer service award alongside some of your other live agents. Can you talk to me a little bit about that? Yeah, that's right. Why and what she did that was so great?

So our company, we do have like quarterly awards that we give out to exceptional employees. And look, we have fantastic employees here. And Claire by no means is a replacement of the actual people that we have. But it was kind of an exciting moment to share.

So we did give Claire a customer service award in front of our whole company at our town hall. And she did give like a little acceptance speech and was very willing and excited to receive that. And so it was interesting. We actually even had to train her to give the acceptance speech because she is so, I'd say, humble and focused on solving the customer issue.

And so we're like, okay, you need to, in training her, you need to actually, you know, say thank you and talk a little bit more about what this means for the company and for you. and such. So it was a fun, fun little side activity we did. Great.

Now, how are employees and customers reacting to Claire? And do you have any other results? You said you're able to do things a lot faster, but any other good KPIs or results that you've seen from using? Yeah, for sure.

So I think the biggest thing is our CSAT, which is interesting. You'd expect your CSAT to be lower, potentially as a result of having these tools. But this last year, our customer satisfaction score was the highest we've ever had in the history of our company. Now, that is a result of having tools like Claire, and it's also a result of some of our fantastic people in our support organization and their hard work.

So it's a collective effort, but we did have the highest CSAT. And I think a big reason for that is customers don't have to wait on hold ever to get answers to their questions. So, you know, if you call any major hotel or airline or any kind of support group, you're going to wait on hold, even if it's great support. And with Claire, she answers immediately and most cases resolves your issues.

And if not, she's gathering that information and getting it ready to transfer to that live agent. So it's a positive experience that way. Other things that we're doing is we're actually working on embedding Claire into some of onboarding some of our training some of our even sales types of processes and so it really expands further than even just customer support as well it's kind of becoming a part of our product so to say and the whole customer journey and then what does the handoff look like because i know voice ai agents are great but when you have to hand off to the actual live agent is there time built in Or is there information that gets sent to the live agent so they are aware?

Just again, for that customer experience that the customer doesn't have to repeat themselves, even though they've had a whole conversation. Can you talk to me a little bit about the strategy around the handoff? The ideal handoff, and this is still a little in the works, but the ideal handoff is going to be if a customer says, hey, can I talk to a live agent? I have this issue.

Or if it's a scenario that they do, Claire will say, absolutely. now the wait time is three minutes let's gather this information I'll put you in that queue and then we hand you over as soon as that person available and so then she able the customers like not frustrated to talk to her now and actually give her a chance and see maybe she can answer their questions maybe not But the customer's in the queue still getting to that live person. Now today what it is, is she would gather that information and then it records it all on our ticketing system for that customer.

And the agent, as soon as that agent picks up the phone on the customer, they've got the initial information around that customer, what was discussed with Claire, and a lot of the initial troubleshooting kind of scenarios that does take a lot of time in your normal support call. And so we're finding that it is saving time for our live agents, and they're able to be a little bit more effective on getting straight to resolving their issue rather than trying to have to troubleshoot what the issue is.

so what have you learned so far about integrating voice ai in the customer service organization what have been some challenges some best practices just other lessons learned yeah you know i think there's two things so one i think we assume ai is this magical tool that's going to solve all our issues as soon as i turn it on people will adopt but it's more of a psychology experiment than it is even on the tech side. So our biggest limitation isn't implementing the tech. It's how do we help train our customers to be accustomed to using this tool and to trust it and to gain confidence in it, as well as how do we help our internal team members recognize that this tool will help them and not threaten or harm them.

And to this point, it has been a help and an aid for our company. And then I think another aspect is there is just there's certain stigmas around different AI tools that within higher levels of the company, you have to build a strong business case. You have to understand what is this actually going to cost and create a strong model. and it's hard today in this world we have you know it cost us many tokens to do this but we had to really be tight on our pilot to translate what it would actually cost based on and what the impact of that cost would be so that we could build a strong business case to actually prove out the viability of this as well because it's not worth it if it's going to cost us several million and just increase expenses.

And, you know, it does help the customer experience, but like at some point we can't, you know, do it if it's going to hurt us. Of course. Yeah. No, that's a good, it's a balance that companies are trying to figure out because it's not easy proven until you do it.

Right. So it's, it could be a tough balance. So do you have any advice for our audience who are starting to explore voice AI agents or anything a nugget of wisdom maybe to share? Yeah, I think one of the biggest things I would say is when you're training an AI agent, you have to look at it as if you're onboarding a new employee.

And a lot of times companies are poor at onboarding their new employees. They just assume that the employee will figure it out over time. And then they blame the AI agent when it doesn't do something right or it hallucinates but the reality is is they don't have documented processes and so i would say the key to success here is to be really tight on the job that you need it to be done and to just iterate every day make a change see the data see how it's it's adding value or hurting yourself and you're just making tweaks it's just kind of a fun little science experiment every day we're seeing how we can make it better for our customer experience that's great well nathan thank you so much it's been a pleasure and uh i really appreciate you sharing your story it's it's a it's a challenge that a lot of companies are facing so it's nice to hear from a company that's that's kind of seeing it work so thank you again for sharing absolutely thank you lids Appreciate it.