CXChronicles Podcast with John Spottiswood 234
Adrian (00:00:05) - Guys, thanks so much for listening to another episode of the CXChronicles podcast. Super excited for today's show, guys, we have John Spottiswood, chief operating and data officer at Jerry. John, welcome to the CXC podcast, my friend. Thank you, Adrian, happy to be here. So, guys, today, as you know, it's slightly different show than normal. John and I are sitting here in Caesars Forum in Las Vegas at the live CCW event, so we've got a lot of things going on behind us right now.
Adrian (00:00:28) - We're literally the main stage right over there, so, but we're excited to have John on the show. We're excited to wrap through a bunch of awesome things that that he's been able to do: building his customer focus, business leader journey. So, John, let's start off today shows. We start off all these episodes.
Adrian (00:00:41) - Man, take a couple minutes just to tell us: how did you kind of get into- not just the role that you're in today with Jerry, but, yeah, how did you kind of get into this space of life where you're, you're, you're thinking about customer interactions, customer experiences both in real life and in the digital front? How'd you kind of get into this whole, this whole world you're in today? Alright, I'll give you the quick overview.
John (00:01:01) - So I spent my first four years out of business school basically running stores at Toys“R”Us. So that's where I really learned the importance of customer experience. Because you know, honestly, customer service is just a small part of it, right? Yeah, story. If the products are all stocked and the prices are right on the shelf and you stock good products, you don't have lines at the register. People don't go to customer service, right.
John (00:01:25) - But when they do come to customer service, you've got some kind of a problem, yeah, and it's an opportunity to learn, yeah, to fix your operations and also to really impress those customers to make sure they want to come back to you, right?
Adrian (00:01:35) - Yeah.
John (00:01:36) - But then you know my earliest internet startup experience: when I left Toys“R”Us, I went to matchcom as president and at that time we were- we were- mostly digital, you know, as an early internet company. So we did all of our support through email and at the time we used a very early email sort of customer support management system called Kana, which I think eventually was acquired by Verint, maybe 10 years ago. But from that point forward I've always owned customer experience as part of my overall responsibilities.
John (00:02:11) - And today, you know, 30 years later, I'm at Jerry. We're the largest digital insurance and loan broker in the country, and I would say in the past five years there's been more innovation on the customer experience front than I really saw in the first 20 years. You know, the first 20 years it was all about how do you get the customer to your human, you know, a support agent as efficiently as possible, and, and now there's just so much more with, with, first, conversational AI and now generative AI in collaboration with humans.
Adrian (00:02:42) - Love it, man. That is, by the way, John, a topic that we are already hearing. A lot is talking about AI. Everyone's talking about all the different ways that they can leverage technology to really kind of continue to push their teams and their businesses ahead, and that's that's no surprise. But, John, I'd love to kind of start off by jumping into just so, give us like the 60 to 60 to 120 second on Jerry, before we get into your team, before we get into some of the incredible things that you and your team are building and working on every day.
Adrian (00:03:12) - Give us a high level for what Jerry is, what your customer base looks like, and give folks listeners that are familiar with Jerry. Give them a high level for kind of what, what y'all do and everything with that.
John (00:03:21) - I will do that. So I mean, ultimately, the reason we focus so much on automating our customer service is because it was a necessity for us. Okay, Jerry is the largest and fastest growing digital broker of car insurance and car loans. Awesome in the United States.
Adrian (00:03:38) - A great name. Do you mind me asking where the name comes from? We saw.
John (00:03:47) - So basically, we got to the point where, in late 2022, we're seeing almost 300,000 conversations with our customers every month, 5.5 million users of our mobile app and just a lot of messages coming in. Yeah, so you know our team could not keep up. We cannot retire fast enough, train fast enough. Insurance is very complicated. We operate in 50 states plus the District of Columbia.
John (00:04:10) - Yeah, so it was a horrendous amount of work, and so we, you know, we started early on experimenting with Google, Dialogflow and other technologies that we thought had the promise to enable us to automate the service portion. Yeah, because part of what we did is we automated everything else, so you can go through from shopping to buying online in a matter of minutes with Jerry. That's all automated. But as soon as you want to talk to somebody, yeah, it became this manual process of human right. So you know we were. We were truly at the point where
John (00:04:44) - a customer would chat or text with us and they wouldn't get a response back for 12 to 18 hours. And that's not the experience you expect when you're living in the digital world. So for us, it was really when ChatGPT launched in November of 2022 that we said this potentially could change everything. Put a Tiger team on that, our CTO was involved. Within a couple of months, I got involved in that Tiger team as well as we started to see potential. And we ended up launching our virtual agent system in April, on April 30th of 2020. So a little over a year.
Adrian (00:05:26) - That's awesome. That's fast though, by the way. Yeah, yeah, it was fast. I mean, for our listeners, like John is skating over that one. That is fast to be able to come up with that type of solution, manage that type or leverage that type of solution inside that type of volume, because there's always gonna be things that are breaking that you're changing and iterating, of course. So that's quick to be able to kind of figure that entire flow out and then put it into distribution within a year. That's pretty impressive, man.
John (00:05:52) - Yeah, it was the key to that, I think, for folks that are interested in doing this is getting the CTO and some of our early, kind of most flexible, knowledgeable engineers involved without a lot of design. It was mostly just product and engineering, iterating quickly to figure out how to build.
Adrian (00:06:13) - Love it, love it, man. That's huge. So let's talk about your team a little bit. So now we got the sense for sort of what Jerry is and what you guys are doing, what type of space you're helping to sort of refine and optimize every day. Let's talk a little bit about some of the guys and guests you've got on your team or some of the different teams, the different departments, the different roles that are inside of Fuel and Jerry on a daily basis.
John (00:06:31) - Yeah, so the teams involved in our customer experience, they're obviously, we have a product team and engineering team that are specifically working on the Propelix platform. So Propelix is what we now call this virtual agent system platform that we built. And then within that, there's a new role called a prompt engineer. You can think of this as a virtual agent trainer, right?
John (00:07:01) - And they're basically each day figuring out how do we make our existing virtual agents better and what maybe new virtual agents with different specialties do we need to launch to be able to answer questions that we either currently aren't answering, meaning they're getting escalated to humans, or we're not answering as well as we could possibly be answering. And then of course, we still have very valuable sales and service representatives that answer both digital and voice inbound questions.
John (00:07:34) - And those folks are an integral part both of providing a great customer experience, but also helping the other parts of the team deliver on the automation promise, right? So they're the ones who see what gets escalated. They think about how do I solve this problem? And then they share with the prompt engineers and the product managers how that could potentially be enabled through the virtual agent system.
Adrian (00:07:59) - I love that. And just, John, I got to call this out for some of our listeners that are still in some of the early building modes of getting their businesses off the ground and they're thinking about how to get better the way you just said, which is taking things straight from the user, straight from the customer, straight from the guy or the gal that's using the product or the service that you're putting out there into the world. Man, this is, guys, why we talk about this notion of building a voice of customer task force or a voice of customer group.
Adrian (00:08:22) - John just called it almost like a tiger team in a way. But guys, this is why we're always talking about on the show, the faster you can get your engineers, your product builders, some of your solutions engineers, the guys and gals that are building on the back end of this stuff, you get them close to those customer conversations, you get them close to that customer feedback, solutioning gets expedited much quicker. Your ability to increase velocity and get it shit done faster improves.
Adrian (00:08:47) - So this is one of the easiest things you guys can be thinking about with your own businesses as you scale. You don't have to wait to get to Jerry's size to be able to do some of this stuff. You can start flexing some of those muscles now and really kind of starting to get good at that.
John (00:08:58) - That's right. And to give a sense of how quickly this works is anybody in our service team, if they see an opportunity, can take the conversation, put it into a project management system we use called Asana, create a ticket, it gets alerted automatically in Slack, and one of our product managers will be looking at and prioritizing that issue within a matter of minutes. Yeah, so it's a quick response.
Adrian (00:09:24) - Keep expanding on the Jerry Tax Dex, because you're already getting there. You're talking about Propolix, you're talking about Asana. What are some of the other tools, or what are some of the other solutions? And again, guys, John and I are sitting here at Caesars Forum. We got, what, 250, if not more, hundreds of providers that are out there on the expo floor.
Adrian (00:09:39) - But what are some of the other tools that you guys have had to really make investments, or make bets on, to be able to allow the business to scale, and be able to allow the business to achieve some of the volume that you're achieving with all the customers you're dealing with every single day? It's a great question.
John (00:09:53) - So we, in terms of what the customer interface is with, we use TalkDesk for our voice response system, and we use the IVR within TalkDesk. We use Twilio for our underlying voice architecture, both for the phone numbers that go through TalkDesk, as well as for all of our SMS. And we use a, we built our own chat interface. So in our app, the in-app chat is something that we built on a framework that's kind of an open-source environment, and we like having that control.
John (00:10:28) - But ultimately, if somebody chats or SMSs with us, those messages go through a collaborative customer service platform called Front, which some of your listeners may be familiar with, some of you not. It's been very useful for us. And so when a message comes in via chat or SMS, it gets picked up by the Propelix virtual agent system, which, in 100% of the time, attempts to answer that question, right? And if it can't, which about 7% of the time, it ends up not being able to fully resolve the customer issue, then it gets escalated to a human.
John (00:11:08) - But at that point, if the virtual agent system has gone back and forth, say, two or three times, clarifying with the customer, its problem is asking them to try a solution. They try that, it doesn't work. At that point, when it gets escalated to a human, that whole trail of back and forth is recorded in Front, just as if another agent, like a level one agent, had been doing that. And so it not only helps customers where it doesn't get escalated, but even for those where it does get escalated to a human, it's extremely helpful.
John (00:11:36) - And in general, if you think about Propelix, like what we've done and how we've integrated with our systems, so it doesn't replace a phone system, it doesn't replace a chat system. You still need some method or Twilio for SMS, some method for getting these messages in. But it integrates directly with all those. And in addition, with our own internal systems, for clients, we've integrated with Salesforce, we've integrated with Microsoft Teams, Slack, RingCentral, et cetera. So it's a pretty easy integration on that front.
John (00:12:09) - On the back end, this system, Propelix, needs to have access to your knowledge base articles as well as your, if you want to get to the 93% kind of coverage rate that we have, it's also got to integrate with your customer database. So that's done through API. So the API pulls back any information about the customer, which enables it to answer detailed questions about their policy, about their payments, other things that they're often inbound asking us about. And then on the knowledge base front, we have 50 carriers in 50 states.
John (00:12:48) - You multiply those two together, you've got 2,500. Right, and then different products. Each one has car and home and renter's insurance, motorcycle, some of them. So there are documents that get into details about each of those policies, all of the details. Like if you're late on your payment, what's your grace period before they're gonna cancel your insurance? This information sits in those documents. So part of Propelix is a vector database that is available to us and our clients.
John (00:13:20) - And it's a simple process of getting those documents into the system so that the information in those documents can be used when answering questions for which that information is relevant.
Adrian (00:13:30) - Love it, love it. And it's funny, so a couple things. Number one, it's just like getting to the point to where technology can do some of that, that's a game changer, man. Because think about it, like even on a simple inquiry where you have a customer that's maybe reaching out about automotive insurance or boat insurance or motorcycle insurance, there's a bunch of, like if you're an agent, you think about your traditional type of agent work, there's a bunch of different things you have to search, that you have to pull.
Adrian (00:13:54) - Guys, every one of us has, at some point in our career, we spent some time on the front line doing stuff. You're literally opening multiple tabs, you're trying to do conversation with the customer while you're literally pulling up the, and so to be able to have, and when people keep talking about being scared of AI or scared of some of this, this disruptive technology that's coming. First of all, this disruptive technology has been in the works now for decades.
Adrian (00:14:13) - There's been a number of different companies that have been building this stuff or building the foundation for what it now is. But the bottom line is, it's really about helping humans.
Adrian (00:14:21) - do their jobs better, faster, sharper. And then when you get to John's point here, guys, around like, you think about how do you build incredible customer experiences, incredible employee experiences, you're getting wins on both sides here.
Adrian (00:14:32) - And I think a lot of people, sometimes they need a little help to realize that it's not just about getting the customer information faster, it's about helping the guy or the girl that's helping that customer on the EX side, on the employee experience side, do their job a little bit better, do their job a little bit sharper, add a little bit of help. Nobody likes to be hanging on the phone and like, wondering how to help you get that last bit of information that you can't see.
Adrian (00:14:51) - So like, this is some of the stuff that I think is so, just awesome in terms of it's such a cool space and land and time right now. It's a business of being able to leverage all of these things and leverage all of these tools.
John (00:15:01) - I completely agree. And I think, you know, one thing is, getting your agents and your team to buy into, when the virtual system can do it better on its own, there's no reason why it has to be limited to just helping the agent. Because even by doing it on its own, it will help the agent by removing that, you know, our agents often were picking up calls, like I said, you know, 12 hours late. Now they're typically picking up a call, you know, or a text or a chat within seconds. And so the customer at that point is still happy.
John (00:15:32) - They're getting their response. And you know, when we started actually, because we didn't have 100% confidence, we sent every message to one of our human agents. So the message would come in, the virtual agent would develop its response. It would send that response into front, but we would let the human agent read it and then press send, right? So when we launched on April 30th of 2023, that's what we did.
Adrian (00:15:55) - That's cool.
John (00:15:55) - But by May 2nd, our agents, first of all, like 99.9% of the time, they were just hitting send. And it was our agents who came to us and said, why are you sending this to me?
Adrian (00:16:06) - Just send it to the customer, right?
John (00:16:08) - And so, you know, I think the agents, they'll realize when they can add value and when they can't add value. And there's no reason to force them to be in the loop if they're not adding value, right?
Speaker 4 (00:16:19) - Absolutely, 100% agree.
Adrian (00:16:20) - So John, so some of the stuff that you guys are talking about now that you're building. So you brought a voice bot demo so that we can literally give our listeners a sense, number one, just how good. This is what me and you were talking about the other day, man, like guys, this stuff has gotten good, right? The quality of some of the bot transcriptions and some of the ways that technology can do some of the tasks that John's talking about. He brought an example for us.
Adrian (00:16:43) - Why don't we go ahead and play this real quick so our listeners can get some, give a little context.
John (00:16:47) - Yeah, so we launched our voice version of the Propelix system on February 1st of 2024. So many months after, and we did a lot of iteration on both voice to text and text to voice, but it's now in our call center. Any call that comes in after hours gets picked up by the voice bot. Any call that is a low priority call, which means the customer, based on the characteristics of the customer, the actions they've taken, they're deemed to be a lower priority. So typically this is not someone that has gone very far in the purchase process.
John (00:17:24) - So as a result, their callback time is gonna be delayed. So we pick them up with a virtual agent first, see if we can resolve their issue, and if not, we escalate it. And the example I wanna play for you now is from that. And with no further ado, let me, let me.
Speaker 4 (00:17:43) - Hello? Hi, Angelou.
Speaker 3 (00:17:44) - I don't know if you guys got my information. I see that you've requested close with us. Would you like to hear about the close options available to you? Yes. Great, Daniel. The cheapest car insurance option available to you is with Progressive at a monthly premium of $270.77. This does not include comprehensive and collision coverage. Would you like to know more about this option or hear about others? No, that one's wrong. Sure,... in the area by choosing the quote you'd like and clicking to confirm your rate.
Speaker 3 (00:18:35) - Confirmed rates are typically available within five to 10 minutes of requesting. Simply click on the notification that will be sent to your phone in the next few minutes to request a confirmed rate. All right, so, that's the
Speaker 4 (00:18:55) - Yes, Daniel, that's correct.
Speaker 3 (00:18:56) - The quote for $270.67 per month with Progressive covers all five of your cars. The 2009 Honda Modest EX, 2003 Toyota Corolla CE, 2007 Toyota Scion GT, and 2009 Honda Fit. This option does not include comprehensive and collision coverage.
Speaker 5 (00:19:11) - All right, perfect.
John (00:19:16) - So, in that instance, she goes on to close the sale with the customer, and the things that you can see from that are, one, obviously, she handles a pretty difficult accent without any problems.
Speaker 4 (00:19:30) - A hundred percent.
John (00:19:31) - She's able to pull data from our database, the quote information as well as the vehicle information to confirm with them. And then, finally, she's able to take an action, which is to send a deep link to the page in the app where he can request the final bindable price. So, a lot of good things happening, and that call, again, would have been very long delayed. It's possible when we tried to call him back, he almost for sure would have had to request a call back and might not have been available at the point at which we called back, right?
John (00:20:01) - So, being able to pick this up with a virtual agent, solve the problem, push him back into the app to complete the purchase is just a huge win for us and for our agents.
Adrian (00:20:10) - A hundred percent. I just think one of the biggest parts of that example, and number one, John, thank you for sharing that with us, but it's just like, guys, as businesses achieve a certain level of scale and as you start getting into the world of hundreds of thousands or millions of inbound touches, guess what?
Adrian (00:20:30) - This is where you need to leverage technology and you need to be able to have almost like a break wall, if you will, almost a break wall where before the waves kind of come crashing all the way in, they get slowed down a little bit or they get filtered, right? That slowdown or that filtration or that compartmentalization, whatever you want to call it, it literally makes your ability to give the customer that much sharper of an experience. It's a game changer. By the way, in terms of this technology too, that sounded great, sounded great.
Adrian (00:20:59) - It was also, for some of our tech nerds that are listening right now, you could hear how quick the speed was for rationalizing or for being able to compute whatever response you had to provide there. Super, super cool stuff. Your team should be proud.
John (00:21:11) - Thank you. So just for those who are the tech nerds, we are- There's a bunch of us on here, John. We obviously built the underlying system. Propelix enables, again, the multi-agent system that gets that detailed answer, pulls the information from the database, et cetera. But we're using other components. So we use DeepGram to do the voice to text. We use 11 labs right now to do text to voice. These are all plug-and-play pieces that you can pull in and out. I'm not sure they'll be the ones we're using in six months, but they're effective for us.
John (00:21:46) - And the beep, beep, beep sound you hear is interesting. Figuring out that sound is like, yeah.
Adrian (00:21:52) - It reminds me of Caddyshack when the pun is going.
Speaker 4 (00:21:56) - Exactly.
John (00:21:57) - We've played with typing sounds and we're testing various things. The really optimal user interface for voice agent interaction is not quite solved, but we're exploring.
Adrian (00:22:10) - But no, that's fantastic. And on top of it, these are some of the things that are going to continue to push the space forward. These are going to be some of the things that are going to continue to be some of the deciding factors of what other solutions or what other type of aggregated solutions you need to plug into the grand plan slightly later on down the line. But this is awesome. John, I want to ask a couple of questions around process.
Adrian (00:22:28) - As the business has scaled, as Jerry has grown as a business, as you guys have brought on more customers, as things have just gotten more complex in general, I guess, how are some of the, what have been some of the ways that you've been able to sort of navigate that change? Here's a simpler way of saying it. As a team's gotten bigger, as a business got more complex, how do you manage all this tribal knowledge? How do you, as things are changing across your team and your tools, and how do you sort of wrangle a process?
Adrian (00:22:53) - Or how do you sort of curate the types of playbooks or type of knowledge content that's required to have a big team of people sort of know what's expected in their specific subject matter expertise area of the business?
John (00:23:05) - It's great. I think one advantage as you build out these virtual agent systems is you really need to have that process, right? If you don't have a good set of knowledge for your human agents, then you're obviously not going to be able to feed the right knowledge to the virtual agents, right? So I think getting organized to make our propellant system, our virtual agent system really successful for Jerry also enabled us to pull this information together for our human agents. And that's been important.
John (00:23:41) - But I mean, I think ultimately the key is thinking of this as one system, right? The voice and chat and SMS-based virtual agents have to be viewed as fitting in with your human agents. So you have to think about, how do I escalate? Whom do I escalate to? So one example of something we built that I think is kind of cool, previously, whom do I escalate to?
John (00:24:11) - We had no way for somebody to be able to schedule a call back. Okay, yeah, talk desk doesn't enable that functionality and which is why I don't, because you need to do this a lot in our type of you know how many times have you called in and and and on you? You're often given the ability to request a call back. Yeah, but it's always like: right now, we're the next available, right next available. So the with the propellant system, it's pretty easy. It can already talk to you. So it just, you know, asks you: would you like a call back?
Adrian (00:24:35) - What time?
John (00:24:35) - You know it can I, can I call it? Can we call you back now or is there a better time? And if you say a better time, it asks you what day and then it triggers and puts you into the call back. You, yeah, on the day that you need it. So again, another thing that makes it very efficient for for for our team, because they're not calling back or being, you know, on a call, that's a call back, and then the customer's not there and they're having to leave a voice message or something. Right, it's just start the whole thing 99% of the time.
John (00:25:02) - Now we're connecting with people when we call them back. So I do think that you know, as you embark on a virtual system, it is imperative that you get a little bit organized. If you're disorganized, I mean most, most operation seems- are pretty organized. Yeah, operational leaders- that's one of the reasons they get promoted typically. Yes, but you got to think about that with your virtual system as well. How does it fit in to this system that you're building?
Adrian (00:25:27) - I love it and you know it's funny when I hear John say that guy's like one of the first things I think about is, again, they want to hate on the notion of why it takes time to do some design thinking or why it is imperative to think about customer journey or user journey mapping- what I just heard you say there.
Adrian (00:25:42) - Oftentimes operationally- and I've worked in a lot of different operative types of positions in my career- by laying out those lines in the sand and seeing almost the stratified view of exactly when a customer or users going through a different part of the journey, yeah, then there's this ability to almost do this mirroring effect. Or internally, then you see where different roles of the batons being passed across, different teams or different parts of the guys.
Adrian (00:26:04) - This is why it's so imperative to be able to map some of that stuff so you can actually see it's not just a journey, it's more about the internal side of being able to see your process, parts of managing that journey as it goes.
John (00:26:23) - I think that's exactly right, john.
Adrian (00:26:26) - I'd love to, I'd love to jump into, like that, the fourth, fourth pillar of feedback, lessons of feedback man.
Adrian (00:26:31) - So I just love to hear you kind of spend a couple minutes talking about, like some over your career and working with all these different customers and working all these super cool businesses, maybe one or two you know golden nuggets for our listeners around things that you've thought around how to get really good at not just collecting the managing active on the customer feedback side, but maybe an example on that, one example that and maybe one example on the same question, but in play feedback, just spend a minute you're talking about such customer and a play feedback, great.
John (00:27:00) - Well, what one lesson learned- and we we kind of went in with this mentality so but it's turned out to be really important- is, first of all, don't build an a, you know, a virtual agent system that that locks people into communicating with the virtual agent. Yeah, I hear people the word chat bot I feel like, has become a negative.
Adrian (00:27:25) - Yeah, because people get stuck, you're right, with a chat bot, right, stuck with it, right.
John (00:27:29) - And so from the very beginning we were very strict about if the customer, if you tell the customer to try something and it doesn't work and you don't have anything better for them to try, escalate yeah. And if they say I want to talk to a human- which happens rarely because they don't, most cases don't realize it's not a human, but escalate. And so we ask you, senator, there are a lot of CX nerds like me that sometimes scream: let me talk to a human. Do you have those recordings? Or what? On voice, we do. On voice, we do.
John (00:27:59) - On text, when you guys hear how crazy I sound when I say human, give you human. Yeah, with voice, you do. Because we, with the voice we always start with, this is Jerry's virtual agent. You know, some people don't know virtual agent means yeah, yeah, digital agent, but but most do. And so they'll ask for a human if, if, if they're impatient, but, but.
John (00:28:20) - But in text it just says: you know, Casey is typing and so you don't know for sure that that's a bot and in most cases it's very rare for somebody to say: let me talk to you, man, but it, whether it's responding through chat or responding through voice, you have to escalate immediately, right, make and then. And then there's almost no negative impact at all to a customer experience, right, because they can always get out of it.
John (00:28:42) - And the other thing I would say is: you know, I think it's really important to break down all barriers to getting that customer feedback through to the people. That can change totally underlying system, right. And so what I talked about earlier in terms of making the process very simple: with us it's a sauna and slack. Yes, with you it may be something different, but but not only making it easy for them, but then always showing, getting back to them, like any time.
John (00:29:14) - Part of the reason that we put it in slack is because we want them to see that a product manager has looked at immediately and and that gives them the confidence that they're not wasting their time when they take the time to give feedback. And when I say them, I mean our frontline, totally agents that are getting the feedback from the customer. And sometimes that goes into fixing the virtual agent system. Sometimes it goes into fixing our product, yeah, the app itself.
John (00:29:39) - Sometimes it goes into fixing automation that we have behind the scenes, but that feedback is is really, you know, hugely important. Yeah, and the other thing I would say is: this is just my personal opinion, but.
John (00:29:55) - I'm here at this conference and I'm seeing a lot of vendors of voice and just CX in general. And there's two issues that I'm seeing with the way that they're implementing generative AI. One is many of them are making it a black box. So you can do certain things within their system, but it's mostly their team working for you to try to build out these virtual agents or bots. And if that happens, and you don't have visibility into essentially the code behind those agents, then one, you're locked into that, right?
John (00:30:35) - So you can't leave, you know, I don't want to beat on five, nine, but you can't leave them if, yeah, or you can leave them, but you can't take your intelligent virtual agents with you. Right. And then the, so that's probably the main thing is I think for most people, the other thing is your agents. So some of the prompt engineers that we have working at Jerry are former agents, right? They were probably some of our best and call it maybe not necessarily the most customer empathetic, but probably this, you know, say the smartest or fastest.
John (00:31:13) - And they've become a prompt engineers with, with Jerry and they own the iterations on, you know, our payment sub-agent and our, and our policy sub-agent and et cetera. And it's been a really good career track.
Adrian (00:31:27) - I love that.
John (00:31:27) - If you take that and you just, you know, outsource that to a company that's going to... you're missing that connection to, this actually is a plus for my career, working interactively with these agents because I'm training them and who's better to train them than your own best agents, right?
Speaker 4 (00:31:44) - Absolutely. A hundred percent.
Adrian (00:31:45) - And the part, the part that you made about to just making a separate track out of it.
Speaker 4 (00:31:49) - Yeah.
Adrian (00:31:49) - There's the EX piece right there, giving a guy or gal that was awesome at a certain thing and letting them be part of the, the authoring of what the future of this stuff's going to look like to be able to manage more scale. That's pretty damn cool. That's pretty awesome.
John (00:32:02) - And if you can point to that and say, Hey, you know, Kerry got promoted to do this as opposed to we're hiring some outside to do this. It's just, it helps build the support within your team.
Adrian (00:32:14) - John, this is why I say it all the time in the show, but I truly, truly believe that so many future C level executives in this country anyway, they're going to come straight from customer face and roles, whether it's sales, whether it's customer success, whether it's customer support, whether it's customer contact because of what you just said, where you start to get equipped and you start to get nasty and understanding the good, the bad, the ugly, how to manage all these different things.
Adrian (00:32:37) - Plus then you begin to over as your career kind of expands and you start to begin to become a little bit more, um, um, aware of how the technology can be applied to it. And you're the human directing how that, that technology gets applied to it. Now you're valuable. And then not only that, now you're starting to be able to kind of really push this whole damn thing into the future and you're going to be able to help many different businesses kind of think about what this is going to look like as we get deeper into it.
Adrian (00:32:59) - So John, this is, this has been fantastic before I let you go. My friend, number one, huge. Thank you for joining us in the CX Chronicles pockets has been a blast, uh, before I let you go. Where can people find out more about your team at, uh, at Jerry and propel X and then where can people get in touch with you guys if they have more questions that they want to learn more about the awesome things that you guys are doing today?
John (00:33:18) - Great. So, um, Propelix.ai is our, uh, Propelix website. Jerry.ai is the, uh, the Jerry website. Um, you can reach me on LinkedIn, John Spottiswood, I think I'm the only one, but if not, I'm the only one in, uh, in Palo Alto. Um, and, uh, yeah, I would be, be happy to give you guys any additional feedback or thoughts or ideas. And I'm also can be reached at, uh, John at Propelix.ai. Love it.
Adrian (00:33:47) - Hey, John, my last question, I'm going to throw the throw, throw your curveball here, but a mini firearm, a blackjack or a lot, I would say blackjack, but then, um, are you, are you pushing or are you holding? Or are you, are you taking another one at 15 when a dealer, when the house has got 16?
John (00:34:08) - Uh, I'm holding.
Adrian (00:34:09) - I like it. For sure. I like that.
John (00:34:11) - Otherwise I wouldn't be playing blackjack.
Adrian (00:34:13) - I'd be playing roulette. This is why he's doing his job. So no, I'm just kidding with you, but seriously, John, this is awesome. I'm so appreciative you taking the time to do this and, uh, I hope you have a blast at the rest of CCW. Thanks Adrian.
John (00:34:24) - Appreciate all you do for this industry.
Adrian (00:34:26) - Thank you. Cheers.
