4 Ways Virtual Agents Are Automating the Contact Center
Brian: Today, we are talking about the four ways that AI-powered virtual agents automate the contact center. I don’t think this is…something everybody knows is that IVRs are limited on their automation capabilities. That’s just the way that it’s been in a pre-AI era or a pre-cloud era. That’s also why most contact centers have an overreliance on live agents for the most routine call types and processes that really don’t require much complex critical thinking.
So, with the introduction of AI, I know that you hear a lot about it on the landscape, now clearly, we’re one of the vendors in that space that are pushing that category and the delivery of that AI over the cloud. For the first time, organizations can get past the limitations of their IVRs. They can automate more conversations than ever before, and here’s a key point, and actually improve the CX when it’s implemented the right way and deliver a perfectly trained agent experience.
So, today’s webinar is going to show you exactly that. It’s going to share some real-world examples of four ways that leading companies are using AI-powered virtual agents in practical ways and using them today. So, with me is Tom Lewis. For 13 years, he was a partner at Deloitte Consulting where he led the customer service advisory practice advising Fortune 500 clients on, I guess, Tom, I guess that would be everything: contact center strategy, technology, operations, customer experience, you name it. He’s currently the CEO at SmartAction. Tom, good to have you with us today.
Tom: Thanks so much, Brian.
Brian: So, we are slated for 30 minutes, but as you have questions, I ask that you just type them into our Q&A box, address it to all panelists. We’ll answer some of those as we go along and we will move to a full Q&A at the bottom of the hour for anything that we missed. So, in front of you is the kind of like obligatory corporate overview slide. I’m not a fan of long company overview slides, so I’ll make it quick. SmartAction started as an AI research company back in 2002. So, we’re not the new kids on the block when it comes to AI development, machine learning, natural language processing. And where we fit is with companies like yourselves who are already using a contact center platform or IVR, might be a Genesys, an Avaya, a Nice, inContact, a Five9, so on, and they lean on us to expand self-service automation in those environments.
So, the bottom of the screen, I don’t need to name them, you can see our awards on screen, but if you do want further evidence beyond just what the analyst community says about us, maybe you want to know what your peers think of us, you can just simply Google, “Gartner Peer Insights” and see who is number one in virtual customer assistance among the likes of the IBM Watson’s of the world. I just needed to get that shot in. So, this is just to set up at least a little bit of street cred about who might be talking.
In front of you is a really simple diagram. We’re calling it “Yesterday’s Contact Center.” You can think of that as just being a pre-AI pre-cloud contact center. And when I show you what’s different about today’s contact center, the biggest change is this introduction of AI-powered virtual agent. So instead of just having simple touch-tone automation or even directed dialogue from your IVR then everything else goes to a pool of live agents, there’s now this pool of virtual agents that can automate some of the calls and chats that used to be handled by your live agents. And Tom, I’ve teed this up for you, the reason behind this shift to more self-service, it isn’t just the AI technology itself but it’s the delivery of it over the cloud.
Tom: Yeah. And you might wonder what the big deal about that is and what the big deal is that it is offering the seamless integration into any, and I mean any architecture. So, no hardware required from our clients and nothing changes except that instead of routing some of the calls to live agents into those queues, you’re now routing calls to virtual agents where they either complete the call or transfer it to a live agent with a screen pop of all of that gathered data so the human can pick up the call right from there.
Brian: Now, voice automation, I mean, it’s been around a while. It’s not anything that we would call new but up to this point it just hasn’t been very good. I remember just a few years ago we were all frustrated with Siri as a voice interface.
Tom: Yeah, and we were really hating on IVRs, but now that’s mostly gone by the wayside in a very short period of time. Everyone is accustomed to the voice to machine interface on their phone, you were talking about Siri, and in their home. And now, they expect it from their contact center and they expect it to be just as good. Now that speech recognition has gotten so good, it’s opened up a new world of cognitive capabilities to do something intelligent with what was heard because very often, the best human to machine interface is the voice.
Brian: Well, I might as well just preempt this question. I guarantee it’s going to come up, we hear it all the time. Does this mean that live agents are going to go away and they will all be replaced by virtual agents, and if so, when?
Tom: No. No, no, I don’t think anytime soon. In fact, I don’t think anybody’s forecasting that agents are going away. But think of it this way, the emergence of virtual agents are upskilling agents to roles that require complex critical thinking or empathy or persuasion. Those are pure human emotions. And the call types and chats getting automated right now are the ones that are routine in nature. So, it just so happens that the majority of calls in the contact centers are repetitive routine in nature and there’s nothing new in that part of the approach to this sort of topic, but it’s now offering a much better CX at a lower price point which means the contact center can actually do it now instead of just talking about it. And we have many customers who offload more than half of their call and chat volume with virtual agents, and this is providing a tremendous relief considering the constraints around agent retention these days.
Brian: Well, when you talk about over half of call volume deflection occurring through virtual agents, I mean, that’s a big number but I’m just going to guess that’s not where most customers start or I guess maybe, do they?
Tom: Well, most customers are probably smart. Start small. And I can say that of our customers because we can allow them to start small, kick the tires, demonstrate success, and then expand to other call types or chats. It’s a very easy and low-risk approach, and frankly, a good example is from world’s largest pizza chain, and I should say, I hope as you mentioned at the top everybody’s enjoying their pizza from Pizza Hut. We are working with this large pizza chain right now to automate their rapid reorder. And so, in other words, we’re allowing return repeat customers to order the same thing they did last time. And apparently, it turns out that a lot of customers simply reorder the same pizza every time and it makes sense. I do that. I got a pepperoni pizza. And so all other custom food orders go to a live agent for now but the next phase will be handle daily specials after specials. The roadmap is to handle food orders.
So, as you can see, although virtual agents are only handling a portion of their call volume today, these simple repetitive things particularly the repeat reorders, it’s not in a too distant future. You may not order pizza or food for that matter from anything but a virtual agent.
Brian: So, then I guess really instead of this big bang implementation from the get-go, it really is just then about identifying the place to start.
Tom: Exactly. And the key collaboration we do with clients is identify that place to start. We’re looking for two to three call types or chats that are the most obvious candidates for AI automation, and that will deliver an immediate ROI.
Brian: Okay. So, you said most obvious call types and chats and what do you mean by that?
Tom: Well, I suppose by most obvious, I mean the call types where the live agent almost feels like a robot doing the job anyway. It’s generally not requiring complex thinking or troubleshooting. Great example, Office Depot has a straightforward return process. They’re not using virtual agents for everything they do, but they’ve identified returns as the best place to start with virtual agents. The ROI is a no-brainer, and the customer experience in that lane is as good as a live agent.
Brian: So, we’re talking about simple call types and chats as the obvious place to start. Now, I should mention, I know that you also work with a lot of customers about solving very complex multi-turn conversations with AI-powered virtual agents, and that’s very different than a, “Simple call type or chat.”
Tom: Right. And that gets later on down the roadmap for a lot of our clients, and it’s where the real power of AI is being realized, and a good example is the AAA Motor Clubs where virtual agents handle the entirety of emergency roadside assistance. So, imagine this. It’s a very complex process. It means the natural language greeting to capture one of several open intents as to why they might be calling, find out their location using GPS, finding and dispatching the nearest tow, checking in with a tow truck operator on location, and getting the latest ETA, and then checking back with the member or the driver to confirm that the ETA still works.
Brian: Well, in this arena of complex, I know that you’re also doing some complex stuff with one of the top two insurers in the world. That might be worth mentioning.
Tom: Yeah, that’s a good example actually. They use virtual agents with their body shops who are doing…well, I suppose the body shops call when they’ve got a damaged vehicle and they want to find out what’s covered on a policy and get authorization to do the work. And this is a conversation where as I refer to it as the happy path, has 17 back and forth or turns between the caller and the system, and each turn presents a fork in the road to a different outcome depending upon the answer given and how that matches against data in the customer record.
And one of the most difficult things to capturing is these long alphanumeric sequences such as the policy numbers and doing it as fast as the caller can say them, and frankly, that’s hard even for a human to do and not possible obviously with touchtone. But an AI-powered virtual agent can do the whole call in about three minutes which is about the same time as a live agent, the difference of course is about a tenth of the cost.
Brian: So, Tom, before we jump into the content, I think it’s worth just kind of setting the table here with a quick definition of a virtual agent or essentially just answering what can it do?
Tom: Well, I suppose to be clear, it’s not talking about simple chatbots. We’re talking about a cloud-based AI brain centralized core capability that is connected to the same data that live agents are seeing, and it has the ability to read and record data into the databases just like a live agent does, has the ability to recognize natural language, extract the intent from the caller over the phone or chat or text, and navigate these multi-turn conversations, and in some cases, even predict why someone might be calling or treating them differently based on who they are when they called.
Brian: So basically, all the tools then to mimic live agent behavior, so to speak.
Tom: Yeah, exactly, but with the confines of defined business rules that act as these guardrails for virtual agents. So, it stays in the lane where we know that it’ll provide an awesome customer experience, as good or better than a live agent, and the best part is that you only need to train that virtual agent once and then deploy it in each of the different channels as opposed to all the training, having to do with agent retention and all that kind of stuff.
Brian: Now, you introduced this word, guardrails for the virtual agent to keep it in its lane so it’s going to do what it’s good at. I think everyone listening in is saying, “Well, hey Tom. What happens when the call goes outside that lane?”
Tom: Yeah, and it’s very simple. Then it’s merely routed to a live agent with a screen pop on the agent desktop with all the data that was collected up to that point so the agent could pick up where the virtual agent left off. And at least at that point, you’ve saved some call handle time and certainly protected that customer experience and avoided customer frustration.
Brian: So, any organization here that has yet to transform even just one call type or chat to AI automation and do more self-service, I know that right now, all of this is kind of a great big black box. And what we want to do is open that black box by first getting into the first of four ways that organizations are using virtual agents right now. Tom, can you just help us first by sharing this concept of natural language intent capture and examples of who is using it?
Tom: Sure. One of the benefits of AI virtual agent solution is that you have natural language processing technology available at your disposal. And it means the understanding of the intent from an open-ended question instead of making the caller sit there and listen to a long list of single word commands or what buttons to press and that sort of thing. So, a good example of this is one of the largest hospitality chains in Vegas and they had so many places that they could route calls that they were having live agents do it because you can imagine the number of menu choices they would have. So, they replaced their live agents answering the phone with an AI virtual agent that simply asked, “How can I help you today?” And depending upon the property, the solution needed to be capable of understanding the intent for upwards of each property, 10 to 15 different end points or route points before routing or taking the caller down into next level of questions.
Brian: Well, let me just pause you there for a second. When you say 10 to 15 intents, I mean, that’s a lot of intents to handle considering all the ways that someone can ask for something at a hotel especially a big hotel. I think that those listening in, I haven’t seen the question pop up, but I’ll just ask it. Our listeners just might be interested in knowing what kind of data was needed for that kind of rollout.
Tom: Yeah. And to be clear, it’s 10 to 15 different end points, the number of intents that could represent that could be huge. So, in this case, a few hundred live agent call recordings were needed to understand how they got to each end point and that was enough data to identify the vernacular, build out the language acoustic model, and then tune the intent capture with our AI engine. And keep in mind that for each intent, we might be listening to hundreds of different things.
Brian: So, is this usually what is required then for most customers to get started if they want to introduce more self-service in their contact center?
Tom: I think it depends on the complexity, but generally, no. I think our centralized natural language understanding engine already has hundreds or thousands actually of grammars programmed into it. And since we support self-service across so many different industries and call types, it’s rare for us to run into a call type we’re not already supporting either directly or have the building blocks for it based on that centralized brain. And so, it’s usually just a matter of customization to augment the pieces we don’t already have, that are unique for that specific customer. And granted, every customer is going to have uniqueness but the amount of effort there is usually not dramatic.
Brian: Now, Tom, I don’t have this example on screen but I know that you are also doing something similar in this area of natural language intent capture with AAA. You mentioned them just a little bit earlier.
Tom: Right. And AAA, they’re greeted by the club’s own IVR system and it’s got traditional touchtone options. If they chose roadside assistance and sort of depends on which club, press 3 for roadside assistance, they’re routed to us where they might need help with a whole bunch of things.
Brian: And I imagine that whole bunch of things, I mean, I can think of, needing a jump, a flat tire, needing a tow and fuel, I guess that kind of thing, Tom.
Tom: Yeah, yeah, yeah. If you were trying to do this with your IVR, you’d have to list all those things, long menu of possibilities to choose from, and we simply ask, “How can we help you?” And before we even ask that question, however, we’re already using their caller ID to pull up their membership to ensure that their account is active, that they have enough credits or tows. We know we’re in the clear to proceed with that call because we have the information we need and that’s an example of a guardrail I was referring to earlier.
Brian: Yeah, I know. If I was listening into that question, I would ask, what happens if the system can’t match the caller ID to their account?
Tom: Sure. We do the same thing as a live agent and authenticate them with their home address to match the account, and that’s actually a good example of what we call our golden rule. The golden rule is however a live agent handles the call is also how the virtual agent should be handling it. So, in that case, let’s go get the address if we can’t do it on their caller ID.
Brian: Okay. So, some of this begs the question, well, what happens if the virtual agent doesn’t understand intent from that open-ended question?
Tom: Yeah. If a virtual agent doesn’t capture the intent immediately, we might suggest a handful of the most common intents and go into a more directed dialogue fashion. Something like, “I’m sorry I didn’t get that. Do you have a flat tire, need a jump, need a tow, or something else?” And even though we’re listening for more intents in what was listed, the idea is we’re making suggestions to the caller on how to interact with the system.
Brian: Okay. So, if you don’t capture intent, then what?
Tom: Yeah. We transfer the call immediately to a live agent, and again, screen pop, give them the data that we’ve gathered so far and allow them to pick up the call from there.
Brian: Now, I think it maybe might be worth mentioning is there anything that makes capturing intent especially difficult or challenging in this case?
Tom: There is, and I think this is part of our secret sauce. I can think of probably three things that come off on top of the head. One is accents. We get that question a lot. Another one is background noise. You can imagine that from a roadside assistance perspective, and the other thing that many people don’t realize is the limitation of the audio quality of the United States phone system versus say high-definition audio device like the Amazon Alexa or even Siri. And so, we’ve got some great technical answers to all those challenges through our proprietary platform, I suppose, which I can go into if anyone’s curious at a later time.
Brian: So, Tom, I’m going to have to back you up to the large hotel-casino answer you gave. A question came in from Mary about that. She said that in some of these cases, it almost sounds like the virtual agent is outright replacing the IVR. So, the question is, should listeners think of virtual agents in that context?
Tom: Short answer is no. I always say if you’re happy with your existing IVR, keep it. If you’re still taking a lot of calls with live agents, then that’s where the virtual agent automation can help. And probably a good example of that is in retail with TechStyle Fashion Group, you may not have heard of them but you’ve probably heard some of their brands like Fabletics, ShoeDazzle, JustFab, they have an in-contact IVR for the initial prompting and routing which works great and it’s economical. But then they send the calls to us for many of the activities that their live agents were previously handling like member services type questions. And keep in mind, this is an AI-powered virtual agent, so the channel doesn’t really matter. And in TechStyle’s case, it’s able to offer all of that same kind of self-service both in voice and in chat.
Brian: Okay. So, thanks, Tom. You just actually alluded to the second way organizations are using virtual agents and it’s in this area we called front-end data gathering, that routine part on the front-end in many calls that can chew up your expensive live agent handle time just because it is so repetitive. And earlier, you mentioned how we authenticate and then pass along the screen pop to the live agent if it turns out the call type should be handled by a human instead of a virtual agent. What are some of the top use cases or, I mean, maybe you can just share a couple use cases and what customers are saying.
Tom: Well, I think one good example is in the medical space. And for those not familiar with it, authentication to even help somebody can take a large portion of the average handle time due to HIPAA compliance. What that means is that companies have to triangulate around three different data points among say six to eight available in their database in order to make sure it’s really the patient. So, social security number, date of birth, that sort of thing. In the case of J&B Medical Supply, this was taking their live agents three minutes on average and they brought in the virtual agents to handle the authentication before passing the call to the live agents and were able to offload all of that time and money and then continued on down this roadmap of another 20 or 30 different self-service applications.
Brian: Okay. So, let’s say beyond J&B Med and I know you’re also doing some really interesting stuff here with the second largest appliance manufacturer in the world, I know that you were speaking at a conference with them recently.
Tom: Yeah. And in fact, we capture all sorts of stuff for them: customer name, product name, model number, serial number, date of purchase, that sort of thing. And so, six months after we went live, the client presented at the execs in the KNOW Conference in Santa Monica, this was this past fall, and showed that both the virtual agent and a live agent were doing the same task in the same amount of time granted, with one major difference, the virtual agent was one-tenth the cost and that was just the upfront data gathering part of that call.
Brian: So, Tom, the third way that virtual agents are being used is likely the most obvious which is fully automating and fully containing certain call types, the high-volume repetitive routine calls that come into the contact center both over the phone or chat. Maybe you can share just a couple examples of what customers are doing here.
Tom: Sure. A great place to start on that one is probably back to the pizza example. When we greet the caller, we trigger a system call to their back-end database using the caller ID to pull up as much information as we can. And one of the things we’re looking for is to see if they’ve ordered within the last two months. And if that’s a yes, we provide a predictive experience and change the greeting and tailor the greeting for that knowing that there’s a high probability that they want to order the same thing that they ordered last time.
Brian: But before the virtual agent even offers the reorder, I imagine that there are some certain business rules that must be met within the data.
Tom: That’s right. We’re looking for a combination of, I don’t know, four or five different things. The order having occurred previously in the last say 2 months or 60 days, that the price of that order still is within sort of a tolerance of today’s price, the store that they ordered from still open, items from the previous order are available and in stock, and then making sure that if we deliver the pizza, that it’s going to get there in the next 90 minutes or so.
Brian: Okay. So, then what if it doesn’t fit then all those parameters?
Tom: Then the call is transferred to a live agent and if it meets those parameters, then it’s fully contained within the virtual agent and a pizza pie is on its way to a happy customer. The key to doing this without sort of sacrificing an ounce of customer experience, however, is by identifying the scope for the virtual agent where you know it’s going to perform as good or better than a live agent, and the way you create those perfect lanes for virtual agents is by using the customer data to define the business rules or guardrails. So, the pizza one’s a great example of that.
Brian: Yeah. So, I’m going to encourage our audience, we’re moving into the fourth and final way. If you are sitting there with any questions, please just type them into the Q&A box and once we get to the bottom of the hour here, we will catch all of those. Now, Tom, the fourth way organizations are using virtual agents, it’s in this area of outbound calls and text.
Tom: Right. A large part of our overall business actually is outbound, and I see this is often an untapped opportunity for many companies because it often requires a lot of human effort but we’ve got a great solution for it. And sometimes, when we go in and talk to clients, they’re not even thinking about the opportunities here. And let me give you an example. So, Penske Truck Rental, they used to use agents as well as outbound robo calls and texts to remind people who are going to pick up a truck. And so, the problem is that oftentimes, the customer needs to change the reservation and when you get a robo call, you can’t handle that.
So that meant that they had to hang up and then make another call into the call center and it’s just not a great customer experience but by using an AI-powered virtual agent, that is connected to the backend data. They can confirm that the pickup’s right, they can reschedule it if they need to, they can change the location, vehicle, all that kind stuff. It’s a fantastic customer experience. In fact, we have a lot of customers who use us for scheduling. It’s one of the things we do really, really well, and one of the largest automobile manufacturers, for example, uses our virtual agents for recalls and scheduling of service appointments at the dealerships.
Brian: So, we’re just about to move into Q&A and we’re nearing here just the bottom of the hour, three minutes away. And now, I don’t want to scare anyone with this slide that’s on screen thinking that we’re running into some type of product pitch here at this point. But Tom, I think perhaps you can just think of this maybe as being your two- to three-minute elevator pitch, just showing with the audience simply why are organizations choosing SmartAction for AI-enhanced self-service?
Tom: Sure. I’ll do a three-story building elevator pitch, so not long. When we ask our clients why they choose us, it’s invariably because we made it less hard than anyone to deploy conversational AI automation and I would say we made it easy. But let’s be honest, creating a customer service experience that’s fantastic with a machine is generally not easy and that’s why we say we make it less hard. Less hard for our client’s customers to self-serve, less hard for our clients to make the transition to automation, and that means that we start with a very small, low-risk, very simple upfront investment, and getting exactly what they need very quickly. And as I mentioned, since we’ve built so many apps across every industry, we rarely start from scratch with any client because all the building blocks are already there in our centralized AI core.
Brian: So, I think one of the most important points we made here is that SmartAction is not just providing the proprietary AI technology, the software, but is actually providing all the services around it to run it too, essentially automation as a service might be the best way to think of it where organizations, they can just outsource the whole thing.
Tom: Yeah, nothing makes life less hard than that. So, case in point, we do not sell our license or software. All this IP, we’ve provided all the services needed along with the technology to transform to virtual agents. We have a team and we haven’t talked about this yet, but we have a team of world-class CX experts who live and breathe this stuff and it’s basically a process of perpetual improvement, and it’s included in the price of what we do. So, we’re working day in and day out to drive CX experience and the best result possible over time. Even though we started as an AI research company and have received tons of awards and recognitions around our AI and been doing that for almost 20 years now, when we think about ourselves, we think in terms of being more a CX company than an AI company.
Brian: Yeah. And if I could just tie a bow on that, the point should be made that the whole thing, the technology of the services, it’s all included together in one flat per minute usage rate. So, the pricing for that matter just couldn’t be any simpler or more affordable, I guess really less hard.
Tom: Or less hard, that’s right.
Brian: There you go. So, regarding next steps there on screen for our audience, most start right here by engaging us for free AI-readiness assessment. You might say, “What’s that?” Well, it’s simply just discovering the call types and chats in your contact center, and after looking at the data, understanding your business rules, understanding the nature, then we can very quickly identify what’s perfect for AI automation and what’s not, and then outline a business case of what that would mean for your business from an ROI standpoint.
Tom: Yeah, and I probably should point out that it’s a very consultative approach. Based on your intro of me and spending all my years in consulting, that probably doesn’t surprise people, but when we examine the call types and the data, what we’re looking for is where virtual agents can provide an effortless experience for your customers and return a strong ROI, essentially a self-funding business case. And we’re very keen and focused on, what’s the problem we’re trying to solve here?
Brian: So, if you are interested, you can contact us straightaway. You see the email address on screen, email@example.com. Just share your availability, we’re happy to jump on the phone with you and start a little bit of that process of discovery and just help you be a little bit more knowledgeable about your environment to know what can and can’t be automated without sacrificing an ounce of CX. So, with that said, some great questions here have come in. So, without further ado, we might as well jump right into it. Those that had a hard stop, we thank you for joining, but we will stay on as long as we have questions and we have a few here to tackle, Tom.
This first one is coming in from Samantha. It has to do with customer satisfaction, just trying to get her head around how your customers are measuring customer satisfaction? Do you see them weighing the difference between what was handled by a live agent and what’s happened by a virtual agent? What are some of your customers seeing here?
Tom: Yeah, and I think that we maybe have a great opportunity to talk to some of our customers about that because here’s the riddle. We’re handling the virtual agent piece. They’re handling the live agent piece. The real customer experience is the combination of those two things. And so, the only way to really get at that data is to run those surveys from the customers or from our clients’ perspective and they do that. And so, what they have found and they report back to us and tell us pretty consistently that when the engagement is entirely handled by the virtual agent, they’re actually…their customer satisfaction scores are even higher than their live agent ones. When it’s a combination of the two, it depends on often how the live agent picked up the call from there. The agent made them repeat things that they’d already done in the virtual agent, then not so great, but if they’re literally picking up the call and moving it along, then they get great scores as well. And so, doing those kind of surveys is part of what our clients typically do.
We do have survey capabilities and we do often ask at the end of our call, “Was it easy to do what you needed to do today?” But that’s only part of the picture.
Brian: So, Tom, some questions here just around integration. Dave is asking, do we integrate with some of the most modern latest contact center as a service software platforms? Another here is, they’re using Avaya, do we integrate with Avaya? Another one here related to everything…all of their contact center ecosystem is on-prem and if we’re cloud, how do we integrate with that on-prem? So maybe you just kind of tackle that integration question.
Tom: Sure. So, divide that into two different parts. One is the telephony integration and we’ve yet to have a telephony platform cloud or premise that we couldn’t integrate with. In fact, you mentioned Avaya, so I should also mention we’re DevConnect certified with Avaya as an example. And so that is fairly straightforward. That effort often takes very little time. On the backend data integration standpoint, that can be bulkier but as our engineering group says, they haven’t met an integration of that kind that they didn’t like. And the logic, sort of our business philosophy has been, we know that it’s a challenge for companies to get that data out of their systems.
On the one hand, they’ve exposed it to their website, so good news is we can probably just reconsume the web services that they used for exposing it on the website. That’s good. Beyond that, however, we know that companies have specific ways that they want to expose that data and we are prepared to consume it however. And so, if that’s your back-end ERP system, whether Oracle or SAP, etc., fine, if it’s your CRM system say Salesforce, great, eCommerce, Sterling, on-prem, cloud, it literally doesn’t matter and the more you have those hooks into the system to make it available to us, great. If not, there’s also…maybe that’s part of the journey because some of it is usually exposed and some of it is, you can see the ROI of exposing it over time. So, lots of talk about there. Best example or best way to approach it is to talk about the specific telephony and data integration that the prospect or client has.
Brian: So, on that topic of data, I have a question here on, hey, Tom, I understand how the virtual agent can be reading the customer data, speaking in reference to text-to-speech and relaying that back to the customer but how is it actually recording the data?
Tom: Well, so the example I used earlier on when I was talking about emergency roadside assistance with AAA is actually that exact thing where we’re looking up the member, making sure that they’ve got the right… they’ve got still credits left for tows, that kind of thing. We’re doing that. But then when we record what they need: a tow, a jump, fuel, whatever, and that ticket, if you will, inside of the AAA system gets created, we’re the ones creating that. And so, it could be a one-and-done where we create it and then the tow is dispatched. Great. If that call for whatever reason partway through ends up on an agent desk because the caller asked for an agent, the agent’s able to look at the record inside of their own ticket management system and pull it, and the way they do that is very similar to the last answer which is through the web services that they have available to conduct those kinds of insertions into existing databases.
Brian: Tom, a question from Nathan here and I guess this would be one for you to tackle because I have no idea, is that, do you have any experience with Amazon’s new contact center as far as any integration there?
Tom: So, it depends what part of that. There is sort of lots of pieces to that puzzle but whether… our view of the world is if you’re talking about the technology stack around Lex, we don’t use Lex, we use our own solution. That may be something that somebody is doing for a part of their business and decide to use us for another part of the business. And then the other part is all around AWS and how they’re using that for their data stores. And so again, we can get a lot deeper into that conversation but they’ve got something up that’s working great, use it. I always say from my consulting days, that your job as a call center operator, or manager, whatever it is, to use the tools in your toolbox and so if you’ve got tools that are working, continue to use them. If you’ve got other opportunities, let’s put some new tools in that toolbox, and in this case, the virtual agents we’ve been talking about could be that.
Brian: Well, I think this is a good take-along question too, and that’s the question, is, they’re already doing some self-service automation with their IVR and the question is, does this mean that if we were to use a virtual agent service, then do we lose the automation that we’re already doing? Does that now flip over to the virtual agents? How do two different automation platforms, then how do they coexist or do they not coexist?
Tom: I’ll give you my standard consultants answer which is it depends but I think that what’s implied in the question is they’re happy with what they’ve got so far but they want more and for whatever reason, either functionality, cost, whatever, they can’t extend what they’ve got. Then having the two co-exist is, from our perspective, no problem at all. One of the things from a customer experience standpoint that might be a consideration is that our system sounds fantastic, has a particular voice and cadence to it, and if the existing platform doesn’t have that, we actually are happy to lend our voice so that it all has a unified experience and it’s just sort of the audio files and that kind of thing. Happy to work with a client so they get this unified experience even though they’re keeping their upfront IVR.
I talked about TechStyle was using inContact up front for theirs. I even said the AAAs were doing that as well, it’s not uncommon.
Brian: Okay. So, Tom, I have two questions here left. One is, you were talking about earlier how some customers are using virtual agents for both voice and chat essentially offering self-service to the same call type or chat type and the same experience. The question has to do with the rollout. Are you rolling out voice and chat identically at the same time?
Tom: Generally, no, and it’s not because of anything other than our clients are usually focused on voice. I think that’s where most of the volume is, it’s where the biggest ROI is. In some cases, they want to do chat, chatbots, etc., but they may not even have a human chat function to be the backstop for that. So maybe again, it depends but our general rollout is go for the biggest bang and ROI which is usually the voice panel but happy to quickly follow along and implement with the same business rules integration, conversation flow, etc., with the chat channels. That’s the more typical layout.
Brian: Got it. So, this question from Simone, are the voice and chat, and text and chat experience, is it the same?
Tom: Not 100% but you would expect that because, for example, at certain points in a voice conversation, you may need to confirm that somebody said something specifically whereas you’ve typed it in, you’re not confirming what you heard, you may be confirming that it even makes sense because if somebody’s said a future date that was actually a past date, then you actually validate that. So slightly different user experience from that standpoint, a better one given that it’s tailored for whichever channel we’re talking about. The key here, however, is that the same business rules and the same data integration exist which means that from a vendor management or a rules-based or what have you, is that our clients have essentially one throat to choke and they can get all of that done through one vendor as opposed to talking to their mobile vendor for this and their chat vendor for this and their voice vendor for that and so forth. It simplifies all of that.
Brian: Yeah, so that way you’re not having any islands or silos.
Brian: So, Tom, just last question here, and I know you talked about the integration, this is just really around implementation, what does a typical implementation look like, speed of implementation? Maybe share just a little bit about that.
Tom: Yeah, the key differentiator and disruptive nature of this thing is that so many of the building blocks of capturing information is in this core platform and that means that most of our effort is spent around tailoring the veneer around the outside that is specific for our clients. So, there are situations where we’ve gone live and as soon as two weeks, that’s not the norm, but more typical is two-three months and that’s very typical and very doable. We usually are running faster than our clients can keep up in that pace, so it’s worth exploring based on the scope that the particular client has in mind.
Brian: Good. Well, I don’t see any other questions that have rolled in at this point. So, Tom, maybe you have any closing remarks?
Tom: No, thank you very much, Brian. Great discussion today.
Brian: Good. Well, for everyone who did join in, obviously, we’re grateful for your time. You will be receiving a follow-up email that has a link to the deck that we used today and once the on-demand webinar renders, you’ll receive a separate email around that so that way, you can take that, share it with stakeholders. Obviously, we would love to work with you as a consultant in this area if you’re looking for more self-service. And all you need to do is just simply just reach out to us and we’re happy to engage you. So, with that said, we appreciate everyone’s time and hope that you have a great rest of the day and look forward to chatting with you soon. Thanks. Thank you, Tom.
Tom: Thank you.