AI Is Taking Over Customer Service: Are You Ready?
Gartner predicts 15% of all interactions with customer service will be self-served through conversational AI by 2020. Hear the 4 ways AI-Powered Virtual Agents are taking over customer service in top companies.
What You Will Learn:
- Examples from top CX leaders on how and where they deployed virtual agents
- How contact centers mix virtual agents with live agents
- Key thresholds to determine the perfect fit
- Which organization should “Buy” and which should “Build?”
Chief Marketing Officer,
VP, Product Marketing & Strategy,
AI is Taking Over the Contact Center… Are You Ready?
Brian Morin: Well, good afternoon, everyone. My name is Brian Morin, I’m head of marketing here at SmartAction, and with me is my colleague and cohort and partner in crime, Dan Fox, who heads up Product Marketing and Strategy over here. One of the best in the biz when it comes to this. In fact, Dan, I think you’re joining us from the Medtrade conference today, aren’t you?
Dan Fox: I am. So, if you hear housekeeping knocking on my door…
Brian Morin: There we know, okay. Well, super happy to have everyone who has joined in today. We’re just setting the table here while everyone is coming into the webinar. We hope that you all are settling in with your pizza. We like to do these… I guess we’d call it a virtual lunch-and-learn, so you at least have a snack in front of you. So, even if the content isn’t any good, hey, at least you’re getting a halfway decent lunch out of it.
If for any reason your pizza has not arrived… sometimes they can be held up by security… we have a whole team at Domino’s on the edge of their seat, ready to serve you and uncover status and make sure it gets to you. So, if there’s any issue, just type in your question into the chat box, and someone will answer that.
And also encourage, just as we go along today’s webinar, we like these to be as interactive as possible; of course, we come with set content, but as much as we can make this audience-driven, we appreciate that. So, just chime in with your questions or comments, even, as we go along. You can put that in chat or the Q&A box, and we’ll handle that as we go through.
Today, we are talking about: AI is Taking Over the Contact Center; Are You Ready? Gartner mentions that by the end of… correct me, Dan, if I’m wrong… by the end of 2021, they expect 15% of all customer service interactions to be self-served through AI. Is that correct?
Dan Fox: Exactly.
Brian Morin: When we look across our customer base, of course, we operate conversational AI for more than 100 brands, so we’re at the absolute tip of the spear when it comes to this emerging trend. When we look across our customers now, we see that a little wider. We’re still in the front end of 2020, and those customers of ours, I would say, on average, looking at the whole, offload about 25% of their call volume or chat volume through AI automation to enable self-service.
So, we’re going to share just a little bit and a few examples of what our organization’s doing to be ahead of that curve, to help you right-size your own organization, whether it might be a potential fit. Because as you know, IVRs, simple chat bots, they are limited on their self-service capabilities, that’s just the way it’s been in this pre-AI, pre-cloud era. It’s also why you’ll likely have many cases in over-reliance on live agents, even for the most routine call types or chat types.
But a lot has changed in this landscape. I don’t think I need to… you’ll speak to that emerging trend with anyone. We’re all aware of that. But what we are seeing is we’re seeing organizations get past that choke point of their IVR or their simple chat bot to automate more conversations than ever before, and here’s the big piece of this: not only are you driving those bottom-line savings, but actually improving the customer experience, and I would say this… when done the right way. Because it all comes down to doing it the right way. You can have the greatest AI tech in the world, but unless you know exactly what the right swim lane is and how to use it, and how to put it in the hands of somebody who can design it well, there’s likely going to be trouble ahead.
So, since we do operate conversational AI for as many brands as we do, it kind of gives us a front-row seat to this emerging trend, and we’re watching how companies are implementing our AI, and what we’re doing to operate it for them, and what it means for their business. We’re typically always doing this work behind the scenes, so it’s not very visible, so the idea behind this webinar is just to kind of pull up a front-row seat for you next to us to showcase how AI is taking over contact centers. And in some verticals and some organizations, we really see it taking over.
And sometimes, these conversations are more left to… you’ll want to want conversations with the analyst community who are trying to stay on top of the trends, and they’re communicating with vendors like ourselves who are on the front line doing it. Well, we thought this would just make great content for a webinar.
So, what we can say is, as far as how AI is taking over the contact center, it’s different for every organization, it’s different for every vertical most specifically, but as we begin to kind of peel back the onion of what different businesses are doing in different verticals, maybe that can help jump-start your thinking about how AI can benefit your specific business, simply by understanding how is it benefiting the broader landscape.
We are set for 30 minutes, and if you have questions, as I mentioned, please put them into the Q&A box or chat box. We’ll answer as we go along, and whatever we don’t get to, at the bottom of the hour we will move to a full Q&A, and we’ll see if you can’t stump the product marketing guy here with your questions, and we will hold on with you for as long as we need to.
I know that a number of the folks joining in, you may not be aware of who SmartAction is; we’re not here to talk about SmartAction, we’re here to talk about customers, end users, what they’re doing with AI, but you may want just a little bit of context. Who’s actually doing the talking here? Do we have any street cred? Do we have any stance on thought leadership to even be speaking on the topic that we’re speaking to? So, from that point of view, just so that you know, what we do is we deliver AI-powered virtual agents as a service, it’s an omnichannel service over voice and chat, and most customers, what we find is they’ll start in voice first. That’s just simply because it has the biggest impact to the bottom line, and then scale the same application digitally to chat or text.
Frankly, we do have others do it the other way around, just the most start with voice, and it augments whatever contact center platform you already have in place. It could be a Genesys, a Five9, a NICE inContact, Cisco whatever. What makes us different is that we don’t just provide the conversational AI technology stack and sell you the software licenses, and then push you out to sea with the ship and say, “Good luck. Here’s a couple oars.” We’ve found that rather than sell the tech, we have to wrap that tech with end-to-end CX services.
So, that means that we have a team of specialists who handle everything from design to build to ongoing operations. Conversation with the machines, they’re not easy, it’s not just design, build and poof, you’re done. It’s why you do need a tea of specialists that are either internally on your team, tuning the application week by week, or on our team which is doing that for you. It’s means poring through data analytics, listening to call recordings, meeting with your team weekly on stats and objectives, and constantly chase that friction-less experience.
And so, we’d like to think the approach is working, as our customer reviews have made us the number one-rated solution on Gartner Peer Insights, just to give a plug about that. All right, well, let’s jump in, after we’ve gotten over the overview and a little bit of a shameless plug there.
In front of you is a simple diagram. We’re calling it Yesterday’s Contact Center. You can think of this as being a pre-AI, pre-cloud contact center. You might have a simple chat bot or IVR in place that is doing some rudimentary self-service, but frankly, we would all agree that most of that traffic is going to live agents. Let’s flip to the next screen to talk about, well, what has changed in this emerging AI landscape where it is taking over the contact center? And the biggest change is this introduction of conversational AI as a layer between the customer and the live agent.
As I mentioned, just as you might have been joining and maybe you missed it if you had just joined, in our experience, looking across 100 brands or more that we support, that means intercepting and offloading a good chunk of that call volume or chat volume before it even reaches a live agent. And I might as well just preempt this question, just because this question comes up and we hear it all the time, it’s like, “Does this mean live agents are going away, and they’re all going to be replaced by virtual agents? And if so, when?”
And the answer is no. Certainly not anytime soon. But you can think of it this way, is that the emergence of virtual agents are doing one big, seismic shift in your contact center, is that they are up-scaling the humans to roles that require complex critical thinking. Or judgment, empathy, persuasion. And so, the call types and chats that are getting automated now through conversational AI are the ones that are routine in nature, and as you know, most interactions in most contact centers are repetitive and routine in nature. There’s nothing new about that. But what is new is the advancements in the capability and affordability of conversational AI, which is really finally making transformation accessible.
The best example I can give about this is that we just had a customer last week… you can pull it up and look online, a big e-commerce customer of ours, they won the Stevie Award last week for Best Return on Customer Service Investment. They achieved their ROI in three months, and just to kind of give you an idea of what that looks like, 50K just to get started and purely usage fees after that from an ongoing operation, and what that meant for them is that their first month, they offloaded 25% of their call volume and chat volume… or, in their case, they don’t do chat. They just do call volume.
They offloaded 25% of their call volume, and so, what it meant is they didn’t just go run out to their contact center and say, “Hey, we can fire a bunch of live agents.” They just let the natural course of churn take its place until they right-sized from an agent headcount standpoint. And so, for them, that meant paying for the entirety of the AI project within three months, and runway to a million in operational savings after the first year. So, that just gives you a little bit of context about how rapidly this landscape is evolving for organizations jumping in. So, Dan?
Dan Fox: Sure. I think it’s probably helpful at this point to give an overview to what exactly we’re talking about when we say conversational AI, and I think it’ll really come to light once you hear an example on the next few slides. But just to set the stage, I think a lot of us have interacted with conversational AI in some form or another, whether it’s a rudimentary, really bad IVR of the past, which probably wouldn’t be constituted in how we defined conversational AI today. Or maybe you’ve used Amazon Alexa, or Google Assistant in your home.
When we talk about conversational AI, we’re talking about a system that is fully connected to the same data as your live agents. So, if live agents have the information about all of your customers through our CRM system, if information needs to be pulled from a knowledge base, that information is connected to that virtual agent.
On top of that, being able to navigate complex multi-turn conversations. Where we see conversational AI, it’s not just asking a question, getting a response, but actually being able to engage in a conversation and the format of how that conversation can look from customer to customer. Just as an example, for State Farm, we navigate the 17-turn conversation. Not everything we’re asking is completely open-ended. A lot of it’s trying to figure out specific information, like an alphanumeric number or a date, but at the end of the day, it’s being able to ask a question, get a response, and then making a contextual reply based on the information that you’ve received.
The most important part here is that conversational AI, the initiatives, shouldn’t be around deflecting every call from going to a live agent, as Brian alluded to earlier, but really being able to capture what you can, capture what customers are willing to provide to you, and then making sure that that transition over to a live agent is seamless. And when that live agent picks up the conversation, whether it’s a chat or a call, they have the information that was processed through automation, because I think we all know that horrible experience of giving information in a bad IVR system and then having to repeat the same information to a live agent.
So, what we’re looking at in the future and what we’re looking at with current implementations, actually, is capturing that information, handing it to a live agent, and having that live agent work in a symbiotic relationship with the AI system to carry on conversations.
As I just alluded to, conversational AI should be omnichannel by design; customers don’t think in terms of channels. They often reach out for channels just because they’re like, “Hey, I think I need to chat because it’s right here,” or, “You know what? I need to talk to an agent. I pick up the phone call.” So, you have to build a system that encompasses all of those expectations and carries context throughout.
On top of that, recognition and cognition. This is table stakes these days for conversational AI, but you need to have a system that can not only understand what people say regardless of background noise and caller accents, but really extract what they mean with advanced natural language processing capabilities.
The one overlooked area which we focus on in all of our applications is prediction and personalization because, a lot of times, if you have ever interacted with a system that says, “How can I help you?”, and you don’t exactly know how to word exactly what you mean succinctly, what you can use is actually prediction, and you’ll hear this in some of our demos, but what we mean here is if you have a bill that you just received and you’re calling, you’re probably calling about that bill. If you called two days ago about a product registration, you’re probably calling back because of that same product.
So, being able to use prediction really helps shorten conversations, and I think you’ll hear that in the demos to come.
Brian Morin: Yeah, Dan, I would jump in, a question here from Katie; she wanted to know, what do you suggest when the content that’s being used by her customer service reps, it doesn’t translate into customer-facing verbiage for a bot to present? And Katie, I don’t know that we’ve actually gotten that question before, but you’re right. It is something that we have to peel back all the time because how a bot is going to handle it is certainly going to be different than how an agent would handle it. In a bot experience, we’re certainly taking a very different approach where we are trying to break that conversation down into chunks, and then what we refer to as hand-holding: how can we hold your hand and get it through?
Dan, I don’t know if you have a reaction to that question, as well.
Dan Fox: Yeah, my feedback always goes back… a question like this, goes back to design and designing intelligently, and keeping in mind that the way that customers communicate with an automated system or a virtual agent is different than the way that they interact with a live agent. So, you can’t just copy and paste the scripts that you use for live agents. What you have to do is transform it into something that an AI system would design.
So, our approach is, give us your live agent scripts and we’ll say, “Hey, we don’t want to say this,” or, “We need to ask a question in this way.” Because it is a completely new channel, a new way to interact, and your customers will respond differently to it than they will your live agents.
Brian Morin: Now, I know, everyone joining here, okay, what are the areas where AI is taking over the contact center? Dan, I know you’re going to kind of give us the bigger, broader picture here of the Where, and then we’ll dig into some exact examples.
Dan Fox: Yeah, sure. So, there’s four things that we wanted to cover today, and this is just where we’re seeing a big impact. As Brian referred to on the earlier slides, the worst thing you can do is look at this and say, “Okay, this is all that AI is capable of.” We don’t want to cramp any creativity. The best thing that you can do is look at a process and say, “Hey, is this a potential use case for AI?”, and approach a vendor or a partner to figure out whether it is or not.
But what we wanted to do is just present a few ideas, which, for us… around 75% of our applications involve what we call an intelligent front door, which is greeting the customer, using predictive intent, capturing what they’re looking for, authenticating them, and then routing them to the right place, whether it’s a self-service transaction or to a live agent. The second one we’ll cover is data gathering, a huge use case for AI in really getting your agents away from those redundant data collection tasks. Repetitive inbound conversations, as you’ll see through some customer examples here. And then, outbound calls and texts. This is often overlooked, but it’s essential to keep in mind that while you’re answering a lot of calls to your customers, are there innovative ways that you can actually proactively reach out to your customers to help deflect those inbound calls because you’re providing them information on the spot when they’re asking for it?
Brian Morin: All right. I guess, Dan, you have an example here of an intelligent front door.
Dan Fox: Yeah.
Brian Morin: I’ll try hitting Play, and see if we can make this work.
IVA: Thanks for calling Klein & Company Men’s Attire. In a few words, how can I help you today?
Customer: I ordered a shirt last… few weeks ago, I guess, and I did receive it yesterday but it’s really big. I think the sizing on the shirt’s a little off because I’m normally a large. This definitely doesn’t fit.
IVA: Sorry. I didn’t understand that. You can say things like, “Place a new order,” “Order status,” or, “Change my address.” Please tell me how I can help you.
Customer: Order. Returns?
IVA: Got it. I’ll get you to someone who can help with your return. One moment.
Dan Fox: All right. I think that’s an interesting use case to use because a lot of times, when customers speak in over-running sentences like that, automated systems tend to fail. Any time that customers speak for over four seconds, the ability to capture that probably isn’t high.
What you can do with a system is narrow down the focus for what you’re looking to get them to depending on your business calls, which for many of your organizations may be unique agent skill sets, or a surfacing self-service. So, a lot of what we try to do in design is present options for the customer. If they are calling about a return, then get them to that place. If they’re calling to speak to an agent about an overdue bill, maybe you attempt self-service, or maybe you transfer that right to a live agent.
So, what we have right here are two examples of customers that are using an intelligent front door. For one, TechStyle, who is the parent company of Fabletics and JustFab, if you’ve seen the Kate Hudson ads. And if you call TechStyle today, what you get, rather than getting an IVR system, you get a virtual agent that says, “Hey, how can I help you?” You can say things like X or Y.
What it also does, which is really interesting, is authenticates the customer right out of the gate, whether it’s by recognizing the phone number that they’re calling from or capturing some information about them, and then proactively knowing if they’re calling about a recent order. So, for TechStyle, they send a package every month or a monthly subscription service, and if you’re calling within the three days of a billing cycle, you’re probably calling about that bill.
So, what we are able to say is, “Hey, Sarah Peters, are you calling about that bill that you just received?”, and what we can do is really drive a contextual conversation. It allows customers to utilize self-service as much as possible.
For J&B Medical Supply, we do a similar thing. On top of a ton of self-service transactions, over six, we greet the customer and we authenticate them within a HIPAA-compliant manner. So, as a healthcare provider, what we’re able to do is capture the phone number that they’re calling from, their account information, and then verify information whether it’s the last four of their social, their birth year, their birthdate. And that all depends on what the customer requires for that specific customer.
So, that’s just an example of an intelligent front door. As I mentioned, around 75% of our customers are using an intelligent front door to help drive self-service adoption into those virtual agent transactions.
Brian Morin: Yeah. As a matter of fact, J&B Medical’s a good example because a lot of times, we will see customers start with this as their application. They might include another two to three self-service applications just to start. J&B Medical is one of those who, they just started with the intelligent front door when they first came to us, and now not only do the intelligent front door but have expanded on app by app, and I think that now they’re running 15 different self-service interactions through the system.
So, I’ll jump in on… the second big area we see this AI take over in the contact center certainly is this front-end data gathering, as mentioned before. I’m going to give just a couple really quick examples here that I can pull up for you, and you can see how this might potentially apply to your business. Electrolux, they are the second-largest appliance manufacturer in the world, and when a customer calls in with anything that is… anytime, I should say, a customer calls in, it’s usually service or warranty-related. And so, they need to register the product first. And that’s something that their live agents were having to do on every single call. They would likely end up doing at least 35 product registrations a day, and taking them a little bit more than two minutes to do that.
And so, we found this as an opportunity for the virtual agent, but this is not quite as easy as you might think it is. It’s actually quite complex, because it does involve capturing quite a bit of information. You have to capture the customer name, and doing that purely through voice recognition. Then also capturing the product name. It’s capturing model numbers, it’s capturing serial numbers, it’s capturing date of purchase. So, six months into the implementation, they showed up at an Execs In The Know conference in Santa Monica to present their findings to other customer care executives, and they showed a breakdown of their data showing that within the first six months, the virtual agent was as successful as their live agents in completing the product registration interactions. In fact, they were both averaging about the same amount of time, only seven seconds apart.
But the one main difference was that, for them, in doing their ROI calculation, is that their virtual agent was costing them 1/10th the cost of an agent to do that. And that was only for just the up-front data gathering part of the call and not the other things that the virtual agent’s doing for them.
And second, I will mention this and mention just this quickly, Choice Hotels onscreen is another good example. They have 7,000 different hotel locations, and every reservation starts with the same five or so questions. You have to capture phone number, the location of where they’re traveling, the date, number of rooms, number of occupants, and then after doing about 80% of the reservation, the virtual agent will pass along to the live agent who will then confirm and book.
For them, that meant that all that time that their live agents were doing up-front asking those questions, there was just no need for them to do it, and all the live agent has to do is simply confirm and book.
Now, the second-biggest way we see AI impacting the contact center, this might seem like the most obvious area, but as much as it’s obvious, you may not necessarily… what’s not obvious is how it’s actually applied, and these are the high-volume repetitive interactions that come through the contact center over phone or chat or text. And of course, these are going to be calls that are transactional in nature, and linear nature for the most part; I’m going to give you just one quick example, just a little over a minute, and you can hear…
This is another real, live production call that we’re showcasing here, and this is with DSW, Designer Shoe Warehouse.
IVA: Thank you for calling DSW, Designer Shoe Warehouse. You’ve reached Shoephoria. Am I speaking with Holly Sims?
IVA: To verify your account, please tell me the zip code listed on your account.
IVA: Great. I was able to find your account. What would you like to do today?
Customer: Address change. Address change.
IVA: Are you calling about order number 304?
Customer: I think so.
IVA: Do you want to change your billing address or your shipping address?
Customer: Shipping address.
IVA: What’s the five-digit zip code of the new address?
IVA: Okay. From the zip code, I already know your city and state. Please say just the first line of your shipping address, such as 1234 Main Street.
Customer: 7838 Martin Way East, Unit 17.
IVA: Now I’ll confirm each part of your address. Was the house number 7838?
IVA: Was the street Martin Way?
Customer: Mm-hmm (affirmative). Right.
IVA: Was the apartment or unit number, number 17?
IVA: Thanks. I have your complete street address as 7838 Martin Way East, number-
Brian Morin: Okay. I think that gives just enough to get a little bit of an idea, to showcase what are the really common, repetitive inbound examples that we can showcase here. Onscreen are three really good examples. Just because we’re short on time, I’ll only give one here, and this is the one that we see with AAA. In fact, this is a really good example of where the real power of AI’s being realized, because AAA uses virtual agents to handle the entirety of the emergency roadside assistance process from end to end.
As you can imagine, that is a somewhat complex process. It means natural language, greeting, to capture one of several open-ended intents as to why they might be calling, finding their GPS location using text, SMS; finding and dispatching then the nearest tow, checking in with the tow operator on location and getting their ETA; checking back with the member on that ETA to confirm that it still works. I mean, you can imagine how frustrating it might be if you tried automating this with touch-tone, and in fact, it’s why they never even tried automating it in touch-tone, because if you did, here’s your experience:
You’re stranded on the side of the road, you call in, and now you have to listen through a long menu of options. You know, “Are you out of fuel? Press one. Is your battery dead? Press two.” So, I think that you can see how all of that will work, and with a virtual agent, all of that is circumvented. So, the virtual agent is listening for… I believe it’s seven or eight different intents it can help with, while it’s also listening for other intents that are reasons for live agent transfer that should not be self-served.
For instance, if someone is locked out of their car, that’s not an intent that AAA wants to automate. They want that one being transferred to a live agent. Separately, if the customer data shows that the account is inactive, or they already have three tows in a year, well, then those calls are taken out of automation and transferred to a live agent. Again, it goes back to, it’s not about trying to automate everything with AI. It’s automating the wide swim lane, where you know the virtual agent is going to deliver an experience as good or better than a live agent, and then you’re using business rules or business logic specific to your business, and then customer data about that customer to keep that virtual agent in its lane. When you know it’s forking out of the lane, you know it’s a live agent transfer.
In fact, I’ll just tail off with how Office Depot uses that. They use virtual agents for returns. Now, they’re not using virtual agents for all returns; they’re only using virtual agents for returns where the customer wants a refund because, it turns out, most customers want a refund and that process is transactional and linear in nature. So, in that case, the virtual agent can provide as good of an experience as a live agent, but of course, greatly reducing that cost, and now their live agents are handling calls that are a little more interesting and complex instead of the ones where they almost feel like a robot doing the call.
And Dan, I’ll have you play clean-up here on outbound, and then we will get to our next steps slide and move to Q&A.
Dan Fox: Sure. Around 30% of our traffic today is actually outbound conversations. A lot of the brands that we work with, while they want to respond to customers calling in, they want to create an opportunity to reach out to customers. So, this could be a reminder about an appointment, if you’re a doctor’s office. Nothing’s more frustrating to me than getting a text message from a doctor saying, you know, “Here’s your appointment,” but if I want to reschedule it, I have to call back in.
Some of our customers are actually using text applications or phone calls to reach out to a customer, provide them a notification of an upcoming appointment, and then if they can’t make it, being able to reschedule conversationally right in the application on the channel that they choose. And we’re doing that today with Penske, with Hyundai, and an interesting one, for Brightree in the collections space, where we can actually reach out to customers who have overdue debt, remind them about their balance, and then help them settle that debt on the spot. Whether it’s by setting up a payment arrangement, a payment extension, or giving a reason why they can’t make a payment. Either way, this is a proactive way to get in touch with customers who otherwise may have not called in.
Brian Morin: Good. So, I know that we have some questions that have come in, and we’ll jump to those. I’ll give just a really quick plug here right at the very end, and potential next steps if you want to continue the conversation with us. If I can say this in maybe 30 seconds or less, why customers choose SmartAction.
We call it our Life Less Hard ethos, in the fact that we make it less hard for anyone to transition to automation than anyone else. We wouldn’t be so arrogant to call it easy; there’s nothing about it that’s easy. But what we have done is provided not just the conversational AI technology stack, that it is a proven stack, that you can see our customer reviews have given us a 4.8 out of five rating on Gartner Peer Insights, the top-rated solution out there related to this.
But the biggest element of this actually has to do with our CX team, our end-to-end CX services that we wrap around the conversational AI technology stack. So, that way, you’re getting a whole team. We have seven different teams across seven different CX disciplines because it takes all of that to properly design and build and operate a conversational AI application. You have to remember that these are conversations with machines. It’s not just design, build, and done; it’s an ongoing process to really get to that friction-less experience, what we’re really chasing.
And so, just here at the very end, if you are wanting some next steps, we just find most individuals take one of two paths. One is request a demo. You might want to hear and see what we’re doing for other companies that have the identical interactions over voice or chat that you do, and see what we’re doing in the vertical that you’re in. There’s likely something that’s already pre-built. We have 12 templates, or templates built for 12 different industries.
Or you just may want to jump right in with what we call a free AI readiness assessment. Usually, the biggest initial hurdle is just finding out, where is there a perfect fit for AI? Or, is there a perfect fit for AI? And we can have one of our experts sit down with you, analyze your highest-volume interactions, find out, do these make great candidates for automation? You’re getting the right kind of volume that would give you that no-brainer three-month ROI, or not.
So, that’s what we could do. You see that email@example.com onscreen, and once this webinar’s over, we will send you a copy of the deck so you have it in hand, and once the webinar is rendered, we will send you the on-demand webinar that you can share with stakeholders. You may not be able to get that till tomorrow morning, but once we have it in hand, somebody will share it with you. That way, you can share that with other members on your team, if this is something that you’re kicking the tires on now.
With that said, if you’d have a hard stop, we thank you for joining, and hope to continue the conversation, but for those of you that have questions, and we have a number to get to, we will go ahead and jump in. Dan, I don’t know if you’ve been looking at any of these questions we want to cherry-pick from, or should I just work from the top?
Dan Fox: I can speak to the first one, heavy accents. That’s table stakes for any AI system today. If you can’t recognize speech, well, then you can’t do anything with the text that you’re recognizing. At least our engine is incredibly resilient to both background noise, caller accents, as well as conversational interruptions.
As you heard in that sample, if someone tells you to hold, if someone says to wait, if someone says, “I need to talk to an agent,” being able to recognize anything outside of the normal flow of an application. So, long answer to say yes.
Brian Morin: Let’s see. A question here: “Are you offering tools to do the actual AI?” It sounds like, you know, are we offering some kind of do-it-yourself platform. And then, the follow-up question is, “Or is this services to design and the conversation’s using your existing technology that you have?”
So, it’s not that. We have our own software platform that we use, and rather than expose that platform to an end user to use, we have a team of experts that serves as an extension of your team, tuning and operating that AI application for you. What that really makes us is you get… sometimes, you’ll walk in and stuff like this, when we talk about AI, and you might put us in the technology provider category. And the truth is, we are. We are delivering the end-to-end conversational AI technology stack.
However, what makes us different is that we’re really coming in more as a partner, because we’re not selling the software licenses. We’re selling a subscription based on usage on our network, and then we’re giving you that team to manage it for you, simply because what we have found out is that it’s far more cost-effective for us to deliver that team across different CX disciplines than for you to go out and try to hire your own team to run and manage that application for you. And so, that’s where we find the best fit in the market for us.
Now, you may be a top-four financial firm; maybe that’s not the best approach for you. Maybe you want to hire all of that staff on staff in-house, but for most organizations, it’s just far more cost-effective to outsource this piece of it.
Let’s see, Dan. I have a question here from Aaron, and Aaron, I don’t know, I may need you to repeat the question a little bit better, because I’m not sure if I’ll set this up appropriately for Dan, but he wants to know if anyone is using voice recognition to further validate on your technology.
Dan Fox: I don’t follow the question.
Brian Morin: Yeah. So, Aaron, sorry, we’ll try our best. Maybe you could chime in just a little bit more detail, and we’ll jump in. Dan, this question is from Perry: “How often does the AI need to be tuned?”
Dan Fox: So, in terms of speech recognition, natural language processing, it is an ever-evolving process, though a number-one thing to remember is that conversational AI is not a product, it’s an ongoing technology and a service. There’s no system that you can build, set up, and it’s like, “Great. Here’s my AI,” and let it run.
This needs to be a continual process of tuning because there’s going to be an ever-growing wave of people who are communicating with your AI system and asking questions that you never heard of, never expected. You’re also going to want to build out additional self-service over time. So, anyone who tells you that it’s a set-it-and-forget-it technology is misleading you.
Brian Morin: Yeah. Right. And one thing that you’re always looking for, and we have a team, their entire job is poring through data analytics to find out, what is the number one reason in any given interaction, for instance, for a live agent transfer where it goes outside those swim lanes, for the virtual agent gets transferred over, and identify where and why is that happening, and is that something that we should compensate for with additional data, changes in the conversation flow, in order to contain that. Or is it just an appropriate decision to keep that with the live agent? So, it’s those ongoing conversations week by week.
Dan, this is from Annie: “Does this require a full IT integration with our systems, or should I consider this bolt-on technology? How should I think of this?”
Dan Fox: An easy way to think of it is there’s… An AI system needs two things to run. It needs a data integration and it needs a telephony integration. That data could be from Salesforce, it could be from another homegrown system; as long as you have an API to access that data, you can build a virtual agent.
As long as you can do a SIP or PSDN transfer in a voice environment, or by leveraging an SMS aggregator, or a chat integration, you can handle the telephony and the integration side to communicate. All you need are those two things, so typically, there is an effort involved, but depending on how lightweight your technologies are and the APIs available, it’s likely not a heavy lift.
Brian Morin: Yep. So, Dan, here’s a great question. “Are your competitors offering, like Amazon service on AWS or Salesforce Einstein, I guess we could throw an IBM Watson in, with the others?”
Dan Fox: Yeah, these are completely different technologies. I think they get at the same goal of creating a conversational AI experience. But it depends on if you want to use the tools that they provide to build out your own system.
So, if you are going to build out your own AI system, you’re going to need a conversational designer, an engineer, an integrator… There’s probably about four or five roles that you would need to build your own AI system, so while they may seem attractive at first, if you don’t have the team in place to use those tools, you probably won’t get too far. And that’s really what our approach is for companies that don’t want to be in the business of AI or building their own solution, we do it for you.
It’s a lot like web development. Amazon and Einstein, those would be kind of like a build-it-your-own website like Squarespace or Wix; for most enterprises who need something ready to go, customized, professional, they build a team to support that website, and that’s kind of what you need for conversational AI.
Brian Morin: Sure, yep. So, you can think of those platforms, certainly, as do-it-yourself, and those that want to go that route, we wish you luck, but this is also why we’re delivering ours as a service. And our delivery to the market is to make sure that we’re constantly giving an ever-fluid, best-of-breed conversational AI stack to the market, which is blended in with tech and pieces that are our own proprietary, or others that are other offerings in the marketplace that are from the Amazon set, or from the Google’s, that we stitch in, if we ever find that there’s a use case where those work for us. It’s just about delivering the best technology to the marketplace, and then giving you a team who can work as an extension of your team to do that for you.
“Dan, you mentioned Salesforce; what advantages do you have over an Einstein?” I think that we covered that here already just a little bit in the sense that we’re delivering the same type of capabilities that you could get through an Einstein, but we’re delivering it to you as a service, but of course, we’re giving you a full, unified service experience omnichannel instead of just something purely over chat.
Dan Fox: Yep, exactly.
Brian Morin: Let’s see. Oh, okay, Aaron came back. He said what he meant is, “Can you recognize customers by their voice as a way to authenticate them, in the area of voice biometrics?” And the answer to that is no.
We have had some our customers talk to us about potentially partnering with somebody who does that, and we can certainly make that kind of partnership happen, but we have yet to see a customer implement that. And Dan, I don’t know if you’ve seen anybody using that yet.
Dan Fox: Yeah, I think that the challenge with voice biometrics is that you need a really high repeat caller rate, and I think we’re all seeing call volumes kind of… individual issues arise. There are some verticals where there are high repeat caller rates, maybe like a utility company or a bank. Outside of that, if you don’t have a repeat caller rate of around 30%, the ROI is going to be challenging.
So, while it’s definitely a possibility to integrate it, it’s definitely a possibility to offer it, we haven’t seen too much of a demand based upon the return that you’d expect.
Brian Morin: Bill has got a interesting question here. He says, “It almost sounds like the virtual agent is outright replacing the IVR.” He goes, “Is that true?” And Bill, the answer is, well, in some cases, that is true, and then in other cases, it’s not true. So, it really just depends what you want to do.
If you’re starting with, and you want the intelligent front door capabilities, the ability to greet conversationally, tracked intent, authenticate and route, well, then, you’re right. Any time that it may be routing back to the live agents, routing directly to the live agent… So, we do see this as an outright IVR replacement.
But other cases, it’s sitting behind the IVR for some organization, and one example of that would be the AAA. AAA still uses their touch-tone front end, but if you press two for emergency roadside assistance, well, then you’re transferred to our network with the AI-powered virtual agent to self-serve that interaction.
Dan Fox: And just to hit on the other question, about whether this is our own tool, yeah, 100%. We use our own proprietary speech recognition, natural language processing; if you interact with our system, we use our own technologies. We are a technology company that provides the services needed to make that technology hum.
Brian Morin: “Dan, how difficult,” this is from Amanda Jordan, “To integrate our CRM…” How difficult is it to integrate with her CRM and Genesys platform?
Dan Fox: I mean, both of those companies offer APIs and we can integrate with them. For TechStyle, we integrate with Genesys; for a bunch of others, NICE inContact. We have over 100 customers, so chances are, if you’re using one of the top 20 contact center providers or CRM systems, we’ve done that integration before.
Brian Morin: Let’s see. This from Erica: “Can customers start small and then grow over time, or are these usually big implementations right out of the gate?”
Erica, most of our customers do start small, for no other reason than we allow them to. I know that’s not the case with others. We’ve just found that a lower-risk approach offering, that lower-risk approach lets organizations kick the tires, see how customers interact with the system, get belief in the ROI, and then after that initial success, kind of like J&B Medical did, as Dan was mentioning, they expand from there to other call types and chats. Very kind of low-risk approach.
But as I say that, there are others who certainly go after the big bang implementation. Choice Hotels that we mentioned earlier, the second-largest hotel chain in the world, we went live with them and it was 13 different self-service applications right out of the shoot.
Typically, the key collaboration, that’s why we mentioned the AI readiness assessment, is just doing that identification on, where do you start? What are the two to three call types or chats, that have the right kinds of volumes… the nature of the interaction is one where we know that the virtual agent will perform as good or better than a live agent, and so, when we find those, what I would call, no-brainers, that’s typically where we start.
Dan, any other questions that you see have come in or maybe that we’ve missed?
Dan Fox: I see a question about smart IVR with a visual IVR component. Yeah, that’s definitely an option. As an omnichannel platform, there are a ton of ways that customers can start the conversation, whether it’s through SMS or a smart IVR…
You just have to keep in mind that a lot of people are calling because they picked up the phone, so asking them to go into a different channel to fill out some questions to get to an agent, it’s a little bit more self-serving for the business than it is for the customer. It just comes down to design.
So, the technologies are there, the capabilities are there. How you choose to design it, implement it, we’re happy to have that conversation.
Brian Morin: Good, and Dan, any other questions?
Dan Fox: I think that covers it.
Brian Morin: Okay. Well, for everyone, as mentioned, we will be following up after the webinar with a copy of the presentation, and when the on-demand is rendered, we will get it back to you. Dan, any closing remarks on your side?
Dan Fox: Well, thanks for joining. Hope the pizza was good.
Brian Morin: Yeah, so, we hope that you enjoyed lunch, absolutely, and looking forward to continuing the conversation with you as needed. So, I hope everyone has a great day. Thanks a lot.