Latest / Auto Collabs / Your Dealership AI Is Only 90% Right. That’s the Problem.

Transcript

Your Dealership AI Is Only 90% Right. That’s the Problem. · Auto Collabs

• 0:00
  If I'm gonna have an AI in my dealership and it's gonna be truly agentic and I'm gonna treat it like an employee, I want all of my data to be able to interoperate the same way I'd want all of my software systems to interoperate if I was gonna teach a human employee.
• 0:18
  [upbeat music] This is Auto Collabs. So it's really interesting, these ins and outs lately, because Kyle and I were doing one, and now you and I are doing one, and it just doesn't feel the same without the third amigo.
• 0:32
  You know? Whoever that third amigo is at the moment. Yeah, 'cause, like, we, we can't sing My Little Buttercup at a piano the way the three amigos did. Gosh, we should recreate that. We can't do the [coughs]...
• 0:45
  thing that the amigos- Yeah, yeah... did. Oh my gosh. Which we should incorporate at some point, 'cause it makes sense. It, it should. It should. Well, you know, everybody- But you know, also, though- Hmm?...
• 0:54
  nobody's gonna confuse us as twin brothers like they so often do. [laughs] That's true. That's true. Somebody was really hardcore on that just the other day. Where, where were we? They're like, "Really?
• 1:05
  You're not brothers? You're not brothers?" And, like, we really don't look like brothers. If you'd look at us, like, face-to-face, you're like, "No, you're not brothers." Yeah.
• 1:13
  You- Yeah, but you know, this is like that thing where everyone used to say I look like Seth Rogen. Well, you did- And-... used to look like Seth Rogen. [laughs] You do not look like Seth Rogen anymore. Uh.
• 1:24
  Like, so that, that is actually really interesting, because you don't anymore. You used to. Yeah. You also don't look like Colby Joyner anymore. Hmm, this is true. He, he morphed into a muscular...
• 1:38
  I, I don't know, like, aspiring professional wrestler? Yeah, yeah. I c- I can see it. A luchador. A lu- [laughs]. See, that's the, that's where they separate, right? He's going pro WWE and you're going full luchador.
• 1:50
  Which I'm still waiting for WWE to have a full luchador character. They just haven't yet. Do- No. Do- No? No. Mysterio? Isn't that a... I don't know. Where were we? Where were we?
• 2:01
  I'm the wrong, I'm the wrong person, I'm the wr- wrong person to talk about that. I mean, uh, how are we gonna segue this one? Today, we're [laughs] talking to- Talk about morphing. No. Yeah.
• 2:10
  Well, I mean, AJ McGowan has gone through a bit of a morph over the years.
• 2:13
  Uh, co-founder of AutoVision, um, and then, you know, went into the Reynolds and Reynolds ecosystem with that acquisition, and now is the VP of Research and Development.
• 2:21
  He's one of the most measured thinkers when it comes to what scale actually means and accepting it in a measured state as we take steps and progress.
• 2:31
  And so I'm looking forward to him talking about what he and the Reynolds team has been up to with all of the AI development, with all their investment in even hardware and software.
• 2:40
  So, hope you enjoy this conversation with AJ McGowan. [upbeat music] AJ, thanks for joining us today on Auto Collabs. Good to have you here, man. Great to be here. Thanks for having me. Yeah, absolutely.
• 2:54
  Okay, so look, I, I know we had met before this, but the first time that I feel like I got to actually know you, we're sitting in a lounge in an airport, and you're like- That's a great story already.
• 3:05
  [laughs] "Can I show you something I'm building?" Which is like- [laughs]... that's like my love language. It is his love language. I don't know if you know that. [laughs] He said, "I thought you'd never ask," right?
• 3:15
  I thought you'd never ask. [laughs] Okay. I want you to take a... Because, uh, any engineer, CTO, person that, like- [laughs]... is obsessed with building things, has, like, an origin story.
• 3:27
  The first time they ever built something and they- Mm... saw that thing come to life, whether it was technology or, like, a little toy or something.
• 3:35
  Do you remember the first things you built that were like, "Oh, I can make something," you know? Yeah, I mean, it, it was actually, uh, software.
• 3:44
  Um, and, uh, I was in first grade, and I did a little, like, uh, kind of like a, like a, Metroid kind of game. Um, and- In, in what coding language? What's that? What code? It was Basic. Okay. In Basic, yeah.
• 3:59
  Um, but, uh, yeah, long, many, many moons ago. But that was the first time that, yeah, I fell in love with building stuff, and I definitely went through my, uh, my hardware era, um, of- Oh...
• 4:09
  you know, disassembling every piece of consumer electronics we had in the house. [laughs] Yeah, driving my parents absolutely crazy. Um, but- [laughs]... um, you know, that, that first love was definitely software.
• 4:21
  That's cool. I feel so seen with the hardware era. Like, I, I definitely- Yeah... did that. And then you probably built computers sometime through, like, middle school, high school, college e- r- range, right?
• 4:30
  We all did that- Yep... next, right? [laughs] Yeah, my, my buddy and I in sixth grade actually had a great little business going where, um, do you remember Computer Shopper? You remember that magazine? Oh, yeah.
• 4:40
  Absolutely. Without a doubt. Yeah.
• 4:43
  So, um, we, we came up with this business idea that we'd send our parents' friends out to Best Buy or Circuit City or whatever it was and say, "Let them sell you on whatever you need to get.
• 4:54
  Write down all of the specs that are on the little card." [laughs] "Come back to us. We'll split the difference in cost with you.
• 5:00
  Whatever it is, we'll build it for you, and we'll come to your house, we'll get it set up, and we'll show you, like, how to use Word and all that stuff." That's a business right there. Come on.
• 5:08
  Back when you can make margin on hardware. Yeah. Yeah, exactly. Wow. [laughs] That's so much fun. What's, what's that big, what's that big computer store that they don't have in my, where I live? Oh my gosh.
• 5:18
  They have the magazine now and, but there's actually physical locations. It's cri- driving me crazy. It's not, it's not- You think of that, uh-... grown up enough... Fry's Electronics? No.
• 5:27
  Well- It's like, there's, like, five of them around the country. There's, like, one in- Okay... Houston. Oh my gosh. I'll think of it before this, this day is over. Nevermind.
• 5:34
  And just, Kyle, you don't know this about me. I also built computers. You, stop. I totally built computers. See, now this is, this is the most shocking podcast I've ever been on. [laughs] Mm-hmm.
• 5:45
  And my son does too, actually. He's better at, than, at it than I am.
• 5:47
  But, but we had this conversation earlier today where, where we were talking about the people that adopt new technology and AI in there, the people like Kyle and Ben Hadley, and probably you, where you would happily run the 26.2 miles or whatever it is- Doing all the hard work in Burn It to get to an end that you could have easily gotten to another way but you built it yourself, right?
• 6:11
  Yeah. 'Cause you adopted all the new stuff. And I was like, "I'm, like, a half-marathoner," right? [laughs] I'll, I like the, the computer too. It's like I'll go halfway.
• 6:18
  I'll be like, "I understand why you enjoy the run, but I no longer enjoy the run. I'm gonna let you guys have fun and do it." Um- Yeah. I'm more like a hop in an Uber kinda guy, but, um- [laughs]...
• 6:27
  but yeah, I get the metaphor. What [laughs] Yeah. But, but okay, so, so what... Because this, in my mind, for any technologist, right? This is, like, the most exciting time, you know.
• 6:38
  You just, there's just so many tools that you can pull out at any moment and start operating, thinking, working toward an end goal.
• 6:47
  And with the resources behind you of a company like Reynolds and, and kind of the freedom that you've been giving, given, what are the things that you're trying to wrap your head around or, or build toward or encourage the team to build toward right now that, um, m- maybe you would have never dreamed possible two or three years ago?
• 7:07
  Oh, that is, you know, such a loaded question, uh, because everything has sa- changed so much from two or three years ago. I mean- Yeah.
• 7:14
  You know, the, the big bombshell drop, you know, on your show is we're working a lot on AI. Um, you know, to, [laughs] to nobody's surprise, right? [gasps] No way. Right? Big surprise.
• 7:26
  Uh, that's, I mean, that has been such a fundamental game changer.
• 7:30
  Like, the ability to, you know, be able to leverage the, the sort of curiosity and talent capabilities, but then to become an expert very quickly, um, on sort of the minutia, uh, is just absolutely astonishing.
• 7:46
  I mean, it has changed everything about how I approach technology as a technologist, 'cause I like- Hmm... to dig into details. Um, and to be able to dig into the details and do it in 15 minutes instead- Yeah...
• 7:58
  of in, you know, 15 days is, is wild. I mean, it changes, changes everything. It changes the way that you can think about trying to solve problems. 100%. Yeah. And that's been just absolutely awesome.
• 8:11
  So, so you- I, uh- Go ahead, Kyle. Hold on. I gotta ask this question- I know... 'cause he did 15 days to 15 minutes, so you know what I'm gonna talk about.
• 8:18
  [laughs] I am obsessed with this thesis and, and my buddy, Ben Hadley, we're both obsessed with this, with, like, like- Compression and decompression... decompress- compression, decompression- Knew it was coming...
• 8:26
  right? And- Yeah... how do you manage that? When, uh, y- you know, like, just the technology bridge across the Reynolds ecosystem is already so broad.
• 8:36
  There's already this, like, decompressed knowledge that is in so many people's heads that also needs a level of compression or, in order to understand it, and now we can compress things at the fastest rate ever, right?
• 8:49
  Explain to me all these nodes of access that I possibly can do something with.
• 8:54
  Decom- compress it down to something, and then now I need to decompress that into, like, explaining it to someone else or creating some net new technology.
• 9:03
  How are you handling that within, like, your personal work, but also working with the rest of your team in order to take this insane ability to compress ideas and then decompress them into work, um, without ruining
• 9:19
  the kind of, like, knowledge base or knowledge share or product that is already set? Mm. That's a Yeah. That's a, that's a really, really interesting metaphor. Um, I hadn't heard it put that way, uh, before, so that's...
• 9:32
  I'm gonna chew on that for a while after the call too, probably.
• 9:34
  [laughs] Um, but, you know, I, I think, you know, for us, when we look at, you know, when we look at the entire Reynolds ecosystem, um, what's so interesting is that there, there's such a legacy there of having innovated across all of the, the sort of individual sectors of technology, you know, kind of figuring out- For a long time- Yeah...
• 9:55
  by the way. Yeah. I mean, for a long time, right? You know, first DMS, first n- single name file. You know, stuff that seems like it's a given now, but was truly, truly groundbreaking.
• 10:05
  And so, you know, the, the cool thing for me coming into it, now that we have access to all of these tools, is that, you know, all across the spectrum, um, you know, things are, things are pushing the edge, right?
• 10:19
  In terms of what was possible sort of in each era.
• 10:22
  And so, you know, a lot of ways I feel like it, it's less a, like, compression and decompression and more like, how do we continue to push the envelope in this era of AI?
• 10:32
  You know, how do we take these tools that we've been given, um, and this massive ecosystem that's out there and go and create things that are, that are sort of net new?
• 10:42
  And the ability, like you said earlier, like, with all the, the resources as well as the customer breadth and number of things that we touch, it's, you know, everything that we do is sort of instantly at scale. Um- Mm.
• 10:53
  And in a world of AI, that is amazing, right?
• 10:57
  To have access to, you know, that much information, that many customer touch points, those, those number of interactions, those, you know, nodes of software that exist across the whole thing, uh, make it possible to do things that, you know, me being, you know, kind of an entrepreneur, startup background, it's like, "What do we wanna be when we grow up?"
• 11:16
  Mm. And now we get to think about what are we gonna do next week, and, and it's gotta be all grown up when we get there, right? Sure. Yeah. High scrutiny. That is, that's a great way of saying that.
• 11:23
  High scrutiny, high stakes. Yeah. Fashionable times. Yeah. What is... So you, you just, you all just had your Amplify users conference- That's right...
• 11:35
  um, a, a little bit ago as of, you know, the time we're recording this. Which is... We're, we're just such a big fan of user conferences in general because you get all the practitioners together. Yeah.
• 11:44
  And you get intelligent questions because you have people who are running into the roadblocks, who are thinking of different ways to implement and like, "Oh, could you do this?"
• 11:53
  Which is really the synthesis, like, could we, right? Synthesis for all the products that end up being 'cause somebody asked that question.
• 11:59
  What is your general sentiment for the adoption speed of dealers absorbing all of this new technology?
• 12:05
  'Cause I assume if they're at, if they're at Amplify, they're, they're already on the front curve of it trying new things and willing to adopt. But what's your sentiment now?
• 12:13
  What it, what that feels like now versus what it felt like, you know, say, 24 months ago? Yeah. It's a really, really good question.
• 12:20
  So, you know, as you said, I think w- as you were starting to talk, I was thinking, well, I think we have a little bit of observation bias, you know, both from our dealer base as well as, you know, then in particular the ones that make it to Amplify.
• 12:31
  Mm-hmm. But, you know, they are really pushing that bleeding edge.
• 12:34
  You know, I've got customers in there that are, you know, digging down with their own models and their dealerships and starting to, you know, really think about their internal DAG graph and, you know, those types of things.
• 12:45
  And so we get, at Amplify, we get unbelievably good feedback and questions and enthusiasm, um, about everything that it is that we're doing.
• 12:54
  You know, I think the, the biggest change, I would say, versus 24 months ago, is I think especially amongst, you know, kind of those, um, the, the savvy dealers that are a little bit more tuned in on, you know, what's happening in general, but then in specific, uh, with AI, is that understanding that all of the data needs to work together.
• 13:14
  Mm-hmm.
• 13:15
  Um, you know, I think that, uh, if you rewound 24 months ago, and maybe even, uh, if we were talking 24 months ago, we'd be talking a lot about data cleanliness and, you know, making sure that we had access to clean data, um, and unified data.
• 13:30
  But I think that, I think that in a lot of people's minds, uh, they were s- still thinking about that almost like a reporting function. Mm.
• 13:38
  Like, "Hey, you know, if I'm gonna have access to data, then it needs to, you know, be clean so that if I run this report, I'm getting, I'm able to join on these three things," right? Yep.
• 13:47
  I think people are really starting to grok how important it is that that data is truly unified across CRM or DMS, service, sales, inventory, whatever it is, uh, for the AI to be able to understand it.
• 14:02
  Like we're-- I'm starting to see people make that sort of cognitive leap of, "If I'm gonna have an AI in my dealership and it's gonna be truly agentic and I'm gonna treat it like an employee- Mm-hmm...
• 14:12
  well, I want all of my data to be able to interoperate the same way I'd want all of my software systems to interoperate if I was gonna teach a human employee." Mm-hmm. Yeah.
• 14:23
  Um, and so, you know, I think that there's a lot better understanding of that, um, you know, than there was just a couple years ago. You know what I'm- And the-- Go ahead... finding interesting is...
• 14:33
  Well, first of all, [laughs] I love that you just, like, glossed over two things, like DAG, uh, the, the DAG model- I was gonna ask for an explainer on that one- [laughs]... but he was just going, so. Yeah. What is that?
• 14:44
  So, like, so basically- Teach us in the back of the class... it's a diagram of all your interconnected systems and how they re- they relate to each other, so that you can follow- Yeah...
• 14:50
  the path of, like, what happens, right? Got it. Uh, and then, and then you used the word grok, which is really like i- it's, it's, it's now become a ver- or, or, uh, an, an action. Yeah. Like a verb.
• 15:00
  But it's, it's engineering speak for understanding the data analysis of what's w- what's underneath your, your data structure, right?
• 15:08
  So it's, it's, like we now use it as, like, understanding something, but it's, it's, it's because- I use it as Elon Musk's AI platform. So I think- [laughs]... you're giving- Yeah. But-...
• 15:17
  most people way too much credit. You know, a grok query is gonna, like, query your database and be able to understand it quickly because of the structure of it, right? And
• 15:27
  then you were like, "Okay, so I've been thinking a lot about this, and I'd love to get your thesis on this." But
• 15:33
  we, for the last, I don't know, seven to 10 years, and especially when the CDP started coming to light, was like, everybody's like, "You gotta get your data right. You gotta get your data structures right.
• 15:43
  We, we need normalization across the entire ecosystem." And what I'm starting to find out, you tell me if I'm wrong here, is actually all that's required is some level of data. Because AI is actually relative, right?
• 15:58
  Um, and, and it, it makes assumptions across relative scopes. It can draw conclusions that don't need one-to-one mapping necessarily a lot faster.
• 16:11
  And so my s- my perspective is that even though we've kind of said, like, oh, automotive has, quote unquote, messy data, which there's a lot of definitions around that, but in general, AI can actually handle that a lot better than, like, a marketing automation platform sitting on top of a CDP that requires a very, very specific structure to be able to execute.
• 16:31
  Are you seeing that? And, like, maybe there's a leapfrog moment for us from a technology perspective in auto because of how AI can handle relationship mapping. So I'd, I'd actually make the opposite point.
• 16:44
  So it, it's interesting that, that you say that because I think a lot of people, um, who are working with AI a lot but not building agentic systems necessarily, um, you know, it's, it's very, it's very easy to think that when you look at like, "Hey, I take these three spreadsheets and I give them to Claude, and it'll kind of figure it out and figure out how to interrelate that stuff."
• 17:05
  Um, what, what becomes challenging is when you try to do it at scale, and you try to do it- Mm... with truly autonomous behaviors.
• 17:14
  Because what you end up with is i- if you let the AI figure that out, it becomes a fuzzy match, but it's always gonna be fuzzy, which means it gets very hard to create, you know, deterministic results. Mm.
• 17:27
  Uh, so it's actually, like, like, uh, one way that I, that I put this as we've kind of worked through this progression, you know, ourselves and building out, you know, our systems and building out Ray and all this, um, is just because we can let the AI solve some of these hard problems, like how do you, you know, pattern match between these two things and see the interrelationship, doesn't mean we're gonna get the best result if we give it the hardest possible problem to solve.
• 17:54
  Ah. So if we're trying to get it to tell us, you know, like, "I wanna run this sophisticated workflow," or, "I wanna get this, like, really in-depth analysis," let's not, you know, tie its arm behind its back- Mm...
• 18:06
  from the starting gate to say, "Well, but first you've gotta, like, cross-correlate all of this data and figure out what the interrelationships are, and let's hope that you get it right every time that you try to do that," since they're kind of fundamentally non-deterministic anyway.
• 18:21
  Mm. So, so it's, it, uh, the, the, the leap, I think, is in, uh, in thinking about, like, those marketing systems and stuff, like, you're right, they were rigid.
• 18:32
  Um, and so you had to get it right because it had to conform to a given structure, but they were technically more flexible.
• 18:39
  You just had to have programmers go in and program in the exact flexibility that you wanted- Right. Yeah... and then it would work that way all the time.
• 18:47
  AI can be this sort of, you know, illusionary thing, uh, where it looks like you can just kinda let it do whatever it wants, but when you really try to scale that up and get complex and try to trust it, well, it turns out that, like, the fundamental software engineering still really matters.
• 19:03
  Matters a lot. Mm-hmm. Yeah. And you can use AI to try to, like, "Okay, well, I'm gonna pattern match this, and then I'm gonna, like, shove it into this deterministic funnel," and it's really good at that, right?
• 19:13
  And then you can try to figure out how to use that to make the process easier. Um, but it...
• 19:19
  I, I, I think that that, that understanding of AI and how good it can get you to, like, that 90% mark, um, then becomes a real barrier to taking it from 90% to 100%, because- Yeah, because nobody, nobody wants a 90% answer, right?
• 19:35
  That's right. What... You need an, an answer, right? A fully confident- 'Cause a 90% answer is-... sounding 90% answer is the worst. Well, I mean, AI... Yes, that's what I was gonna say. It's always confident.
• 19:44
  It's very conf- I got on the wrong train once 'cause it was so confident. [laughs] Yeah.
• 19:48
  I was in, I was in New York City, and I was like, I knew I had to make a decision, so I just literally took a picture of the board in Penn Station where, and where all the things are.
• 19:57
  I was like, "Which train do I need to get on?" It was like, "Get on this train. It's departing from track eight. Go now." Literally, it said that. Yep. I was like, "Done," on the train.
• 20:04
  And I was like, "Wait a minute, this isn't the right tr-," and I'm moving. The train's leaving, and I, I'm like, "This is the wrong train. Why did you tell me this one?" You know what it said.
• 20:13
  Everybody knows what it said. I'm so sorry. "My bad." You're right. My bad. "That's on me." Sorry. [laughs] "You're right to call me out," right? [laughs] "I interpreted this little thing.
• 20:20
  Actually, that little symbol meant this, but I, you know, I was in a hurry, so I d-..." Like, that, there was my 90% answer. Oh, well, you interrupted on your day, and I apologize. [laughs] Yeah, yeah, yeah. My bad.
• 20:29
  You're right to call me out. Uh, I know. I can't do anything to you, though, [laughs] right now, but I want to. Um, let me ask you this, uh, as we have our, our closing few minutes here. Sure.
• 20:39
  Y'all have been working so hard on Ray AI, which is your proprietary AI. What does it do now, and what do you hope it... What are you building it to do in 12 months from now?
• 20:48
  Yeah, so, uh, what it does now, and so going back to actually what we were just talking about a second ago, the, the thing that's really unique about Ray, I think, versus, you know, kind of most of what's out there, is that it's not a chatbot, right?
• 21:02
  So in kind of the traditional, you know, first gen of these tools, a chatbot would follow a very deterministic DAG.
• 21:10
  It would just do it in such a way that it kind of was a parlor trick, and it looked like it was thinking. [laughs] Yeah. Ray is, you know, a true software agent.
• 21:19
  Like, we've taught it how to goal seek against whatever the user is asking for versus what our tool set is- Mm... yeah, and to go pursue those goals.
• 21:29
  Um, and so what we're allowing her to do right now, um, is probably the, the better way to say it, because there's a lot that she can do that, you know, when it- Oh, I didn't realize it was a she. All right. I like it.
• 21:40
  [laughs] Like Star Wars Rey. Um. Yeah, okay. Yeah. [laughs] Yeah, there you go. Um, what, what we're allowing her to do right now is to pursue answers to questions.
• 21:49
  Um, so she can look across the whole Reynolds ecosystem using whatever the user's access rights are and say, [clears throat] "I need a little of this from here. I need a little of that from here," [clears throat]
• 22:00
  to answer, you know, simple questions, obviously, like, you know, "Hey, uh, you know, how many Fords do I have?" And she can figure out the right tool and go select it.
• 22:07
  But you can also ask her, like, "Hey, I've got my managers meeting coming up in three hours. Um, who do I need to go yell at?" And she will go and figure out department by department what's going on and pull back- Oh...
• 22:19
  tools and, you know, do, do the whole thing. So, you know, today we're letting her answer questions and do analysis. Where we're heading towards is, um, is allowing her to actually go effect action.
• 22:32
  Um, so- Actually yell at people. Yeah. [laughs] Yeah, I don't know if that's the right use case, but- [laughs] Um, but, uh, hey, actually, it's funny.
• 22:41
  We used that as a, as a test question the other day, and Ray responded back and said, "I'm not gonna tell you who to yell at. I wouldn't recommend that you manage like that." Um- [laughs] Okay. All right...
• 22:50
  but, um, that's probably a real, that's probably a little too much. We don't need the extra commentary, right? So tune that. [laughs] Easy. Stay in your box. Um. Yeah. [laughs] Yeah, exactly. Bad robot. Yeah.
• 22:58
  Um, but, uh, we also- So effect action 12 y- 12 months from now- Yeah... to actually- Next few years... figure out what the goal is and then go set it in motion. And go do that thing. That's right.
• 23:08
  Well, I think so many people have these, like, meeting structures, right? Uh, uh, it, it might be, like, we were level 10 at my stores, right? And, and so everything was rocks and goals and things like that.
• 23:18
  And that, that seems like such a practical thing to feed something that can look across the entire ecosystem and say, "Hey, look, I've got my level 10 meeting. This is the structure. These are our data.
• 23:28
  These is our goals." Mm-hmm. Maybe that's already input somewhere. "Prepare that level 10 meeting, highlight opportunities, and highlight wins from the last week so that I can celebrate with my management team.
• 23:37
  Give me exact data across the entire ecosystem." And that's something that, like, would take, frankly, hours prior to- Oh, yeah... having something at your fingertips like this. Oh, yes. Yeah. Yeah.
• 23:49
  Well, it's so exciting. No, that's, that is. AJ, it's always great to just hear about the, the things the Reynolds team is doing. And, uh, well, you know us. We're, we're big fans, and we love seeing, um, I don't know.
• 24:01
  I guess, I don't know what you... An, an established player just coming out swinging, taking the actual knowledge of the infrastructure and the new technology and putting it all in its place. Um- That's right...
• 24:12
  it was a lot of fun to talk. Next time we're gonna get you to play those guitars that are sitting behind you. If you're just listening, he's got two... He's actually got three. You can't see the third one.
• 24:19
  We saw that beautiful Jaguar in the back. [laughs] There he goes. There it is. You can see the neck. Uh, but thank you as always for joining us here on Auto Eclipse. Thanks for having me, guys.
• 24:27
  [upbeat music] I think at the end of that conversation, he highlighted one of the biggest steps that we are on the precipice of taking
• 24:39
  Just as some dealers, or a lot of dealers actually, are just getting comfortable with the first step. The first step is having an AI that can look across all of our data and give us actionable insights.
• 24:49
  Tell us, "This is what's going on in here. Here's how it pairs together with this other data over here."
• 24:54
  But this next step that is really starting to be the, the dream, the vision board, and I think will probably happen, start to happen in 2027, is the AI actually taking action. Yeah.
• 25:06
  Something that was coming to my mind to this effect is the, the reps over the years that this guy's been putting in. I mean, he starts coding things in the sixth grade. He's developing games.
• 25:16
  [laughs] He's doing all these sorts of things. He's absolutely obsessed with it, and that's, like, a narrative that we haven't really explored a lot when it comes to the AI equation, which is
• 25:28
  there's something to be said of this full immersion into an ecosystem to truly get clarity about its capabilities.
• 25:34
  Like, you could be sitting on the sidelines and being like, "Oh, you know what'd be cool is if I did X, Y, Z with AI." People that are fully immersed in it are like, "We're already doing that." Yeah. You know?
• 25:44
  It's not until you get fully immersed, like the way this guy does.
• 25:47
  And maybe it's 'cause I'm getting old, Paul, but, like, this idea of putting your reps in is really an interesting conversation to me 'cause, like, in a world where we're talking about everything moving so fast in its capabilities- Mm-hmm...
• 25:58
  the idea of there's still something to be said of, like, getting your hands on something and
• 26:04
  playing with it and building with it and- Yeah, I mean, well, you, you are probably one of the, the biggest tinkerers that I know, right?
• 26:12
  And you start to get your reps in across platforms and in many other areas, not in the auto data space, but in many other areas, right? You can see AI taking action for you- Yeah...
• 26:22
  and starting to handle things without you having to handle it. So, um, yeah, you have to get your reps in. You have to get-- You have to tinker, even if it's a little bit, a cursory knowledge, understanding.
• 26:32
  And I think the more people get used to innovating and trying new things, the more, as they say, the game slows down, right? Like, say football- Mm-hmm...
• 26:41
  experienced football players, the play unfolds much slower to them than it does to someone who is new at the game, isn't used to getting the reps in. So, uh, all that being said, Kyle's not here, but we'll still say it.
• 26:52
  On behalf of Kyle Mountsier, the other amigo, on behalf of Michael Cirillo, and myself, thank you for joining us on Auto Collabs. [upbeat music] Welcome to Auto Collabs. Welcome to Auto Collabs.
• 27:09
  [laughs] Why are we recording? Are we rolling yet? Not yet.