Short answer: Anthony Monteiro says vehicle data like year, make, model, trim and mileage is structured data machine learning loves, so they analyze every vehicle listed online to value cars in real time instead of waiting for lagging auction data.
Watch this answer from Episode 25 (starts at 14:40) · Episode 25 notes and transcript
From Episode 25: Car Dealers Are Losing Inventory — AI Is the Only Fix, starting at 14:40.
Jordan Metzner [14:40]: Know, just a quick question before we jump into the news a little bit. Maybe all of this wasn't possible without AI, or maybe AI has, facilitated a lot of significant improvements. Obviously, AI is in the name of the company as well. But it'd be great to hear just a little bit about how, some AI has enabled some, product development features technology that probably wasn't able to be generated before AI existed or before the ability to, like, access these LLMs. I think that, computer vision's been around for years, reading a license plate, that kind of basic y stuff. I'd love to hear a little bit about just how AI has had, like, a major impact on the business.
Anthony Monteiro [15:15]: Sure. It's it's funny. Sometimes I almost wanna take AI out of the name. Right? Because Everybody, oh, I'm AI, and realistically, most aren't. There's some sort of a application that they sit on top of their old legacy stuff and build and say they're an AI company. We started from scratch going into it. My last venture with a startup, we had to lean into AI really heavily because we built a logistics platform of picking and dropping off vehicles and to determine how long it takes to go from point a to point b, and each driver had different characteristics and traffic patterns and time of day and all of these things, AI had to really maximize it or else you would never be able to rendezvous the two endpoints, the a to b, and get the driver to another place. Similar to, what DoorDash and all these other Ubers do. Right? They maximize. They have AI figure out where's the next and closest place that this driver can go to. And so we had to lean into it way before the LLM models that are out there today, but it got us comfortable with, hey.
Anthony Monteiro [16:31]: This is gonna be something that's really gonna help us. The automobile industry and especially what we do with vehicle values, it's very structured data. It's structured. Right? There's a year, make, model, trim, color, options, mileage, right, service dates. There's all these really finite points of data that machine learning loves. Right? It doesn't have to go out to this sprawled place of LLMs and try to pull it back. It's like, okay. I have everything right here. I can like, I've already did the math upfront. Now what do you want me to do? So we were able to take, like, the structured data, and then now go out and be able to see every single vehicle that's listed online for sale, and build algorithms using AI of what's the real value of a vehicle today, right now in that market. Not waiting for the auction data, not waiting for some feed to come in that's normally lagging, and who knows what the vehicle values. It could change in a minute.
Anthony Monteiro [17:28]: Right? Like, there could be some news that completely blows up the value of a car because something happened or something changed it, some economic thing. Well, those happen in real time, and if you can utilize it in real time, then that that number is so accurate. You can't do it without it. You just can't, because you can't process that much information without being able to use AI.
Jordan Metzner [17:53]: Alright. Well, that's I mean, that's obviously a great overview of the business, and it seems like the upside is obviously in your favor as these are kind of historically nontechnical businesses implementing tech technology. So it seems like there's a big upside here. Maybe this is a great pivot to jump into the news, Sam.
