Thought Leaders in Artificial Intelligence: John Price, CEO of Vast (Part 4)

John Price: The other thing we do with the image is, we can determine what’s the make and model of that car just from the image. From there, we can generate all of our analytics about that car. Now you can just take a picture of a car and we can generate a CarStory report for that automobile.

With the help AI, we’re able to use those images to complete the dataset, analyze which images are of value, and to derive merchandising information straight from the images. That’s a big piece of our AI. Another big piece is in search. In big purchase categories, we don’t believe that search is the right metaphor. Drill down is a great metaphor to pick the right TV for your wall.

When you’re trying to make decisions where SKUs are literally unique to just one unit and you’re trying to compare across them, drill-down search is a terrible metaphor. We’ve flipped it on its side and we’ve created a discover approach to cars today. You start giving us a little bit of information and then we use our algorithms to create a set based on all the parameters that would be in the consideration set of what you’re looking for. By doing that, you don’t miss out on things.

That discovery algorithm is extremely powerful when we moved it over to real estate because finding a home is a discovery process. We have a lot of groundbreaking work there. Some of the other cool things that we’ve done is the market analysis. One of the things we can do that is extremely valuable is, we make market predictions. We can look at a piece of inventory and we can predict when that vehicle is going to sell. There’re a lot of applications.

Our predictions are extremely accurate. From there, we can predict at what price it will most likely sell at. On the consumer side, we’ve implemented a full chatbot that  allows you to go through the car buying purchase process with chat. You can go as far as scheduling the appointment for test driving after you’ve discovered the vehicle.

Sramana Mitra: Let’s talk about real estate. I can understand how your large-ticket purchase concept applies to real estate. That extrapolation is reasonably simple. Tell me a bit about who are the buyers, users, and what are the datasets you are drawing from and what kind of applications you’re building on top of that?

John Price: It’s actually an extremely similar problem. It’s a problem with massive amount of inventory. Each SKU is unique. You have that same problem. The other thing that’s similar about automotive and real estate is they’re both perishable. It’s only on the market for a while and then it disappears.

When you map to the fact that it takes you six weeks to six months to make a home purchase decision, the reality of that situation is that when you sit down and get ready to look for your dream home, the likelihood that it is on the market at that moment is less than 10%. The likelihood that it will appear and disappear during that six weeks to six months is greater than 90%. The likelihood that you will find the optimal point is less than 5%. It’s a wonderful application of alerting and AI.

Once you can glean the features, you can statistically get close and find similar homes, then you can start combing inventory and setting alerts on all kinds of things. If you think about the state of the market today, you get an alert when you search for a home. I literally have 25 alerts uniquely related to the home shopping experience.

This segment is part 4 in the series : Thought Leaders in Artificial Intelligence: John Price, CEO of Vast
1 2 3 4 5 6 7