Spyne AI Changes the Rules for Online Merchandising, Conversion Rates, and Communications

Brian Pasch0:00Hi, this is Brian Pasch and welcome to another exciting podcast. It is time that we start to dive in to some of the technology breakthroughs that we're going to be talking about at the Modern Retailing Conference this November in Palm Beach, Florida. And I am so excited to have Sanjay Varnwal. He's the CEO of Spyne. Sanjay Varnwal. And you're going to want to lean into this interview because, well, maybe it's time for you to reconsider what AI with really smart integration and merchandising can do for your dealership. Sanjay, welcome to the podcast.

Sanjay Varnwal0:48Thanks a lot for inviting me on the podcast, Brian, and super delighted to speak with you today.

Brian Pasch0:55So Sanjay, you and I were talking, the platform Spyne, which was started as a merchandising tool for online businesses to give them product-specific, real-time engagement with consumers who are shopping, didn't start out just automotive, but you've been finding tremendous success. And automotive, before we talk about it, what was the problem you were trying to solve with your technology?

Sanjay Varnwal1:32Right, Brian. So I have an interesting question. So one of the core aspects of modern automotive retailing is the way the cars are shown to the consumer. Because most of the transactions that are happening on the dealership side are actually finding its route on online discovery, right? And online discovery on CarGurus, CARFAX, AutoTrader, their websites, social media, all these places, right? So customers, when they are viewing the car, they want to be seeing that they're not only seeing that particular car, but they're also comparing you with the other dealers that are also selling that car, right? So, and you are pitted against maybe a Carvanas of the world or CarMaxes of the world, which are spending like

Sanjay Varnwal2:19tremendous amount of money, resources on setting up those very premium turntable studios, hundreds of locations, right? A small dealer, if you see, they are not having the kind of money or even the space to set up that dedicated studio, right? So there are 70, 80,000 dealers in the country and you don't find them on the equal footing to compete with Carvana and CarMax on that particular front. displaying their merchandising, right? So the idea can be create an easy enough technology that can give the dealer a way to shoot their car right away at their yard. And a lot of digital studios, which are very premium looking, always that car can be transformed to showcase the cars by images, by

Sanjay Varnwal3:02beautiful 360 spins, by AI powered video merchandising, which depicts the car in its true sense and make it, I mean, it is kind of your online shop, right? The better it looks, the better the consumer view about your business would be, right? So that's what we targeted and started targeting like long tail of dealers that cannot afford that kind of studios for their business, right? And that's where we started matching the quality of the outcome with those of like the studios. And we found that traction in the industry.

Brian Pasch3:34Yeah, you know, when I saw your merchandising tech, you know, back in the day, you know, early innovator was SpinCar, right? And there was a lot of focus on, you know, trying to make the car more shoppable, right? More sticky. More sticky. Yeah. But today, you know, when I just laugh at how quickly AI can generate images, I always wonder, like, why, why can't we do a better job with imagery? You know, recently, I'm coaching dealers about website conversion. And wow, the pictures are just, you know, oh, I can't even, you know, I got to be kind to my friends. But man, the, the range of quality, when you're shopping for a car, and the information on the VDPs is really all over the place, right? There's no standard.

Brian Pasch4:27So when I saw what you can do to transform a normal set of photos of a car, to instant 360 spin, looks like they have a million dollar photo studio. Yeah, that was like super cool, meaning the quality was amazing. So dealers who are listening in, and if you haven't seen Spyne's tech, go to their website, check it out. But, you know, Sanjay, that was just the start. We have to get that consumer to start talking with the dealer. And, you know, every dealer who's listening in will agree, hey, great photos, get people interested, maybe a little stickier. But that's just one piece. The other piece is how do I answer their questions about the car?

Brian Pasch5:25What was that second approach to saying, hey, if we got them hooked, let's start a conversation? What was that mindset of being, in a sense, a 24-7 concierge for the dealer?

Sanjay Varnwal5:42Yeah, absolutely. I mean, you, you, you, because you're so deep into the industry, you know where this should naturally act toward, right? So, we got the dealers present their cars beautifully online, on their websites, on the social media places. Now, the next question that came from the dealers is, hey, I am able to present it, but what about engagement? What about the leads that I'm able to generate from here? Because now the next part that we want to attack was, or target was, hey, we have made the shop beautiful. Now, can we make it responsive enough for the lead to be generated and a potential to be booked for the dealer, right?

Sanjay Varnwal6:22So, the next natural step forward was to initiate the conversation part on those cars, right? So, what we did was, we created 24 by 7 human-like agents, which is available on voice, chat, and emails, and deployed it on our merchandising widget itself. So, on the website, when the cars are displayed, and you are interacting with it, you also see, hey, I'm available 24 by 7, the chatbot, the call widget, all of these things. Once you tap there, it basically starts the conversation right away. I mean, 10 p.m. at the night when you are, after your work hours, you are back in your house, browsing different types of cars, you want to get the instant information, detailed information about that car, right?

Sanjay Varnwal7:06And you don't want to wait for the next day for someone from the dealer to call back and then, like, so dealers would lose the opportunity and customers will also lose their interest. That was the moment that we thought that if we can deploy this 24 by 7 agent, which talks like human or chat like human on that particular widget, it will have tremendous impact in converting that particular lead, right? And not only solve the problem for the customers, but also increase the demand generated for the dealer. That is working out pretty well in the industry. Like, we have been piloted it with a few dealers, pretty large dealers from our network, and showing great trends in that right now.

Brian Pasch7:45Yeah, and Sanjay, that's why I want to encourage dealers to come out to the Modern Retailing Conference, because I told Sanjay, look, if you're going to be a part of this conference, you got to come armed with data. So they're working with some of the largest, most named brand dealer groups and individual stores. But I said, look, bring the data. And here's the key. If we understand the first time a consumer is looking at a car, goes to the VDP, the car is the star. They, oh, yeah, that's the car. That's the color I like. That's the interior. And then they shift to the math box.

Brian Pasch8:27How much is this car going to cost? And then they got some other questions. Now, some people might say they'll go start doing payments. But a lot of times, it's clear that there's questions that they have before the payment or the trade, you know, step. So by putting in the communication engine right in with the widget, let's just say the iframe, which represents the vehicle, obviously the attention is there, not in the bottom bar of a website. You know, the car is there. And, of course, people have seen now, like, next generation AI chatbots, email, of course, sending an email form. But what I really said to you is, like, hey, when you come to MRC, let's talk about this voice interaction.

Brian Pasch9:21So your vision is to create a series of vertical agents across the whole automotive retail system. But for this conversation, let's say you had to train these AI voice agents to understand everything about the CARFAX, everything about the vehicle from the manufacturer. Let's talk about that. How successful are your vertical agents sitting on these VDPs able to, when that person at 10 o'clock, 11 o'clock, 7 o'clock in the morning wants to know more about the car, what happens? Is this, are they using, like, a VPN? Are they dialing something from their phone? What does that look like if I'm on my mobile phone and I hit, or on my desktop, and I want to hit, talk to our AI bot?

Sanjay Varnwal10:16Yeah. Yeah, yeah. So, right. I mean, imagine the life of a customer, right? When they are visiting the dealer's website. So they focused on buying, let's say, BMW X5. That's the car that they have, 2022, 2020 model, X5, black color, right? So this is the make, model, hand-time, variety of variations. Now the customer wants to also know the historical information about that car, CARFAX report. Some of the peculiarities about that particular car, whether this car had any kind of accidents or something in the past, right? Or maybe how many hands it exchanged in the past, like some of these information, right? So which is not obvious when you see the VDP page, right?

Sanjay Varnwal11:00So at that moment of time, you see a bot which is available, like which is blinking, that can call me or text me, right? And you call me available, 24 by 7, you tap on it, and there's a human-like bot, just like I'm talking with you. This is the kind of conversation that the bot is having with the customers. It starts with, hey, I see you are interested for this BMW X5, 2020, great choice. Let me know what all questions you have about this particular model that I can help you answer, right? So customers can ask about the historical information of that car, the financing information about that car,

Sanjay Varnwal11:39the variety of aspects, right? So we are integrated with a lot of standard CRM platforms, VinSolutions, DriveCentric, Elead, all of these platforms already. So we are able to pick up the customers' contextual information from those platforms as well, if that customer is already connected. We are integrated with standard inventory provider platforms, such as the vAuto, Vincue, and all these platforms. So we are also able to pick up all the historical information and deeper information about that car, integrated with CARFAX and all. So all these information, because it is an AI, it is way, way more capable to hold all the information and be able to present you in a very logical fashion, all the answers, right?

Sanjay Varnwal12:23So customers can have very intelligent and almost a near completion call with that particular agent, where the agent is able to answer like 80-90% of all the queries that customers could have. Which a human would not be able to, because they will be like going back to the CRM, looking at some information, going back to the VIN, like inventory provider system, looking at some information. And be able to, like, at max arrange a callback or try to provide maybe 20-30% of the information that the customer would want to, which a bot can take this conversation to maybe 70-80% completion.

Brian Pasch12:57Wow. So Sanjay, let's talk about that a little bit. Dealers have, maybe had mixed reaction, you know, in the past with like a digital voice assistant or an AI chatbot. Some dealers are afraid about the AI agent going off the rails, right? What has changed? I mean, dealers who maybe even a year ago tried some AI chatbot or a digital voice assistant that was supposed to be training on the data in the DMS or the CRM. And what if they had maybe a less than awesome experience? What would you tell them now about this time in the marketplace, the way LLMs have evolved, boundaries are training, concepts have evolved?

Brian Pasch13:55What would you tell a dealer who maybe was burnt in the past?

Sanjay Varnwal13:59Yeah, I mean, there are two, three aspects to this particular problem, right? So one is the technology itself was pretty new two, three years back. Like, ChatGPT came in, like, just three years back, right? Three, two, three, two, 2022, it came on. So, but it is seen like a rapid evolution, extremely, extremely rapid evolution. And if you see, like, industries, which is real estate, insurance, health, all of these industries, there are companies which are dominating the first level of inquiry resolution, right? So, and you have, as a dealer, you see, like, you have huge call volumes coming in, or you're sitting on huge data on which you do the outreach is on, right?

Sanjay Varnwal14:42So, humanly, it is not possible to attend the call 24 by 7, or reach out to, like, thousands of data that you have for certain use cases, right? So, there is a need that is emerging. The need is there. There is a burning need of reaching out to those customers who are attending when the humans are not around. So, this is, like, one need, right? In terms of technology, it has evolved at a stage now where there was a problem of, like, bot not sounding human-like enough. That was, like, one of the problems. There was another problem where the pauses were, I mean, the response time was slightly more.

Sanjay Varnwal15:22So, it was making the bot.

Brian Pasch15:24Yeah, I was feeling like it was thinking too long. It was a lag.

Sanjay Varnwal15:26It was a lag, right? So, these things were giving away that we are not speaking to a real human, right? And third was, of course, the kind of integration depth that this kind of a product could require was also not there completely earlier, right? So, I mean, these three problems are kind of getting better and better. We want to attack the use cases which are low-hanging in nature first and get those, like, 80% of the low-end work done for the dealer so that the humans are deployed to some more consultative type of, like, closing kind of process for that particular transaction. And these bots take over a lot of the use cases for the dealer.

Sanjay Varnwal16:09And so, this is one. We also come up with, like, three, four years of experience, very fast-moving, right? So, we have a team of, like, 8,200 odd engineers just on the research. So, there's also a scale that we are looking at for solving this particular problem. Unlike any emerging, like, starting with five to ten engineers, we have got, like, a host of engineers working on a variety of these problems. So, sales inbound is one of the use cases. We are also targeting the service inbound and outbound use cases, which is recall, service inquiries, appointments, some of these.

Brian Pasch16:44Yeah. So, Sanjay, you told me, which was great. I mean, some of the early entrants to DVAs were focusing on those inbound service calls. But, you know, one area is there is money to be made in recalls, but getting people to come in, it's not always so easy. Of course, people who maybe read their emails, but email delivery is a problem from dealer CRM. That's a huge problem. Just task management. I was surprised that one of your early use cases is actually using your vertical AI voice agents to call people about a recall. Now, that's in production now, correct?

Sanjay Varnwal17:30Absolutely, yes.

Brian Pasch17:32So, I really want you to bring some different brands. I'm very interested to see, like, a luxury brand versus, say, an import brand, just to see the success rates. Because I think dealers are not making enough outbound calls to retain their customers, and I understand why, because it's a thankless job in many cases. What are some of the early observations about these outbound recalls? When someone picks up the phone, what percentage of the people are engaging and asking some questions about the recall as compared to, look, I could call as a human to somebody about the recall, and they're going to say, hey, I don't have time right now.

Brian Pasch18:24I'm cooking dinner, click. Okay. But the people who are available, what does an outbound AI trained agent on that particular OEM and that particular model, what does that look like compared to human work?

Sanjay Varnwal18:40So, I mean, two aspects to it, right? The first aspect is, let's say, the OEM has given you this set of data about what the customers who have purchased this car, this particular car, and there's a problem related to airbag or brake, right? So, these are like critical problems that OEM wants to solve, free of cost for the customers, right? So, one is, are you able to convey the urgency and the critical nature of this particular task to the customer?

Brian Pasch19:12That's fair.

Sanjay Varnwal19:12Yeah. Clearly. And the second is, of course, there's so much of data that any human, if they want to do the call, it is such a boring task, right? So, I mean, you see the data is lying with the data, they want to do it, but they find it hard to get it done by a kind of a running continuous process, right? So, these two are the aspects where we created a very smart conversation script to instantly hook that customer. That, hey, I'm calling, there's a problem in your airbag, which can be fatal if you go beyond this particular driving speed, right? And there's a free of cost replacement available at our place where we can give you like pick and drop service as well and get it replaced free of cost, right?

Sanjay Varnwal20:01So, and if you start from a very high, providing the value as well as maybe communicating the urgency of that particular situation pretty well, the customers actually stick to it. And let's say you call about 100 customers, 30 pick up the call. So, our goal is, can we make this customer adhere to that particular conversation, maybe 80 to 90% by creating a script which is extremely, extremely powerful. And so, yeah, that is the direction that we are heading and because it is AI, there's like a no boredom to it, right? So, it can just start and all the customers right away like one by one.

Brian Pasch20:43I love that, you know, because look, I've had airlines call me, you know, like with an automated system and said, hey, there's been a recent change to your flight and, you know, please, you know, press one to speak to an agent. I find that extremely important if I'm running 100 miles an hour or, you know, something critical is happening. I love the idea that you're making it easy for consumers 24-7 and you can choose all channels. It kind of brings up, if you're going to be developing vertical agents eventually for all aspects of sales and service, if they're going to be communicating with you, do you become a CRM or are you always the vision seeing that you'll always just be updating CRMs?

Brian Pasch21:39What does that, what does the vision look like? Because right now, the only way that consumers can talk to dealers is during business hours or in a delayed mode with email.

Sanjay Varnwal21:58Yes, absolutely. So, our entire focus is to build agent, like powerful AI agent that can pick up one job at a time and do it like beautifully well. So, let's say it is sales outbound, recall outbound, maybe some other use cases, maybe related to marketing, updating pricing on the websites or all the marketplaces, right? Maybe after the car is sold, maybe updating it everywhere across CRM. So, maybe identifying what car to buy to source and at what price. So, what we are looking at is we are creating smart job-focused CRM agent that is very good at one particular task and can be integrated with the existing ecosystem of inventory provider systems or CRMs or any other ecosystem plan.

Brian Pasch22:51So, a whole suite of agents then?

Sanjay Varnwal22:54Whole suite of agents, targeting inventory, targeting CRM, targeting e-financing, maybe service, back office operations, all of these, right? And so, the kind of work that human would be doing, we will accentuate it with the agent, which will give the human a dashboard that, hey, I have done these actions. Now, I just need your validation, confirmation or...

Brian Pasch23:17So, integrated with the CRM, but just launching tasks, I know right now I'm in a project where we're taking Urban Science defection data, we're updating the CRM, okay, this car, this person bought a car elsewhere, right? Not from the dealer, we're updating the CRM, but then we're sending out a survey. Well, a very low percentage of people are responding to the email survey.

Sanjay Varnwal23:44Yeah.

Brian Pasch23:45But I bet if I had a vertical agent trained on doing a defection survey, I, you know, this is, you know, Pasch Toyota. And, listen, I know you were talking to our team, but I recently got a notification that you purchased a car elsewhere. Hey, could you just take just a minute with me? I'd like to find out why, right? I bet an AI agent could get 10x, 20x more than... There's only a few percentage are responding. So, we're seeing that they're opening the email, 45% are opening the email, but then you have like a 3% click-through rate, right? You know, so now you're 3% of 40...

Brian Pasch24:29You're like, I need feedback. So, I bet there's be-backs, right? Someone who came in the dealership but didn't buy, is the salesperson really following up? I could see an agent saying, look, with this, I'm Brian, I'm our customer service agent, and we see that you spent time with Tom today in the showroom, but didn't purchase. You know, could you give us some feedback? This could be a true, beautiful thing where... Do you envision this as the CRM could trigger one of your agents to go do a task?

Sanjay Varnwal25:05Absolutely. So, it could be multi-channelized approach, so agent, which is a follow-up agent, let's say you term it as a sales follow-up agent, right? So, whenever a customer does something, so this agent basically is deployed with some rules that for this particular thing, do the voice follow-up, do the text follow-up, do the email follow-up, all of these things. And look at the responses that the customers are, actions that the customers are taking on those things. Let's say you send an email, and that email is opened by the customer. So, now you see, you know that the customer has engaged on that email, though it has not clicked that, clicked on the survey link, but now you can trigger an outbound voice right away.

Brian Pasch25:51I love that. Yeah, the open the emails, the signal, so the call won't seem so out of the blue.

Sanjay Varnwal25:58Yes, yes, and you can capture, like, survey very easily. I mean, you can increase the survey, survey's completion rate by easily, like, but.

Brian Pasch26:06So, Sanjay, we'll have to have a little call about that, because I'm working with some large dealer groups on defection data, and it would be great for MRC if we could train, because I have the questions, right? I have the questions that we want to ask. That could be a big breakthrough. I like that.

Sanjay Varnwal26:27Well, we'd love to, Brian.

Brian Pasch26:29Yeah. So, Sanjay, I love the fact that the auto industry is attracting new innovation. Dealers, for the most part, have only been testing inbound phone agents. I love the fact that you're going to be able to do both. And, in fact, I think maybe more revenue can be generated by the outbound call, because if somebody calls and they don't get what they want, most of the time they're going to call back. But it's the people we're not talking to, but it's the people who didn't call that I'm more interested in, because they're defecting. They're maybe not feeling that the dealership cares about them. So, Sanjay, if a dealer wanted to see your merchandising, the amazing upgrade to photos, and then how your communication platform could be used across texting, phone, and email, what's the best website for them to go to to see it in action?

Sanjay Varnwal27:37I mean, they can check Paragon.

Brian Pasch27:43Paragon Honda?

Sanjay Varnwal27:44Yeah. Yes. And, so, we are right now under, so I'm kind of confidential to not name, like, all these accounts right now.

Brian Pasch27:54But maybe, how about your website, your own corporate website, then?

Sanjay Varnwal27:57Absolutely. You can come on our website and see some of these test bots deployed on the website itself. There will be a follow-up call that you'll expect immediately after you come to our website, and someone will be able to take you through the entire suite of solutions.

Brian Pasch28:14Right. And what is that website address, so everyone knows?

Sanjay Varnwal28:17Spyne.ai. www.spyne.ai. S-P-Y-N-E dot A-I. Yeah.

Brian Pasch28:23So, not spine. It's a funny spelling of spine. So, we like that. We like that. So, S-P-Y-N-E dot A-I. I think, when I look at the conversion rates of websites, Sanjay, you're onto something. Because the, I would say 2%, maybe a national average, 2% of all sessions to a website convert to a lead form. Only 2%. And, of course, some people call, but sometimes call tracking and somebody sees a phone number and it may be a tracking phone number or not. But, we have a very low conversion rate. And I'm wondering if 24-7 they can talk or chat and that the follow-up automation can be more comprehensive, especially on outbound.

Sanjay Varnwal29:26Yeah. I think this is going to unlock greater conversion.

Brian Pasch29:31So, if you would, think about when you come to MRC, bring some dealers who are using the tech and who can talk about it. Because our conferences are about collaboration. And, you know, as a sponsor, you get dealer passes. Bring some dealers who can share some specific data, right? Because dealers are looking. I just had an OEM reach out to me and say, hey, Brian, what's going on with all these companies and who do you recommend? And what I say to people is, look, what's your use case? Have them demonstrate how they solve your use case. Because dealers have different use cases. You know, somebody might be, hey, I just want overflow for service appointments.

Brian Pasch30:20Okay. That's a very simple use case today. There are more exciting use cases that I think generate more revenue. So, that's what I'm going to encourage you, Sanjay. Bring some dealers. Bring the data. Because what I want to do is highlight throughout the conference the data results that are happening. And that's what dealers like. Sanjay, if there was one thing you wanted to tell the dealers listening to today's podcast that we haven't covered, what's one thing you'd like to encourage them with or share your vision with to get them maybe more excited about what Spyne's doing?

Sanjay Varnwal31:00I mean, overall, if you see the AI adoption in the industry, right, it is not linear. It is growing like exponential among the dealers, among all the industry. And you see when something is exponential, right? So, in no time, you don't realize and everyone is using it, right? So, you want to be the first mover on future technology. So, don't view AI as something that will do any kind of damage to your existing operations. It can be a force multiplier, right? So, for the same set of people that you are running your business, let's say you are doing 100 units of revenue, you can increase your top line with the same set of people to maybe 20-30% as well, as big as, right?

Sanjay Varnwal31:47So, if you're AI deployed right way, which is deployed right on services, deployed right on sales, all of these places. And then, the way the ChatGPT, OpenAI have just brought in the kind of possibilities in every industry, right? And you see net positive emerging in most of these industries, like the businesses are gaining so much. Of course, we are yet to see the net outcome impact down the line next few years. But I'm sure that the economic, the way that that is growing, it will have like a positive impact everywhere. And automotive dealership being such a large industry will be at the forefront of it anyway, right?

Sanjay Varnwal32:26And dealers being at the center of it has to adopt, right?

Brian Pasch32:30So, I love that. I love that. It's like, look, when we do searches online, AI summaries. When we have a complex need, you can have AI go through lots of documents and data and make summaries. I think of it, personally, I'm looking at health records, you know, like, hey, look, here's all my blood tests for the last 10 years. Here's my, you know, tell me, are the doctors missing something? Like, that can be done in seconds now. You know what I mean? Like, it's crazy. And we do know that automotive is a little laggard because of the specialization. I want to encourage dealers, look, the early generations of AI and digital voice assistants were great to familiarize the dealers with what's possible.

Brian Pasch33:29And now those platforms have matured and new ones have come in, come out to the Modern Retailing Conference, the who's who of automotive AI tools, voice assistants, and, well, smarter sales processes. Thinkers, we're all going to be there. So it's November 16th, 17th, 18th at the beautiful Eau Resort in Palm Beach, Florida. The conference sells out every year. So I don't need to push people to buy tickets. The conference sells out. So don't wait. So here's my suggestion. As you're listening to this podcast, as we close, go to Spyne, S-P-Y-N-E dot AI, get a demo, ask for your specific use cases, and then come out to MRC.

Brian Pasch34:20Get some experience with tools, see the data, talk with the owners of all these companies, and find a company who's aligning with your vision for modern retailing. You will not be disappointed. Well, Sanjay, I'm excited for you, and I want you to bring the data with you to MRC, and, of course, I'll bring the dealers who want to accelerate their business, forward-thinking people. That's not a problem. The people who watch these podcasts and read my articles are all people who want to get better. So thanks for your time today. Thanks for joining our industry and bringing innovation. And for all the dealers who are listening to this podcast, if this is the first time you're listening, you should know there are hundreds of interviews with automotive leaders, innovators, entrepreneurs, disruptors who are breaking the mold and bringing best of breed tech to automotive retailing.

Brian Pasch35:20And so listen to all the podcasts leading up to the MRC conference, and I hope to see you in November with your team and, well, spend a little time with Sanjay as well and get a chance to see what's up with the team of Spyne engineers who are laser-focused on revolutionizing automotive retail. Sanjay, thanks for your time today. It's been a pleasure.

Sanjay Varnwal35:46Thank you, Brian, for hosting us, and I'm very, very excited to join the MRC conference in Florida.

Brian Pasch35:53Well, it's going to be a great time. So everyone who's listening in, I'll talk to you again on a next podcast episode. Otherwise, I'll see you in Palm Beach in just a few days. Have a great day, everyone.