Thought Leaders in Artificial Intelligence: John Price, CEO of Vast (Part 5)

John Price: With that, we combine lending process, which is the other part of home shopping. Our technology is being embraced very rapidly by the lenders. You look at the mortgage business. They are stuck at the bottom of the real estate funnel. They have to compete on interest rates. What we’ve done is we’ve gone to all the lending institutions that do mortgage origination.

We said, “We will power an entire white label real estate experience for you ran by our Big Data integrated to your lending practices in a way that will provide a far superior home shopping experience to your banking customers.” They’re literally having to compete for their own customers. What we’re saying is they can be at the top of the funnel. You get all the analytics about your home and neighborhood in your home.

We can combine that with other things that you might want us to look for. They can essentially market against the real estate data, get them on the sell side so that they can get them on the buy side to provide them more service. For the real estate in the back end, the problem becomes coordination. I’ve got my loan and my agent. Rarely do the two speak. Now I need coordination. All of these things are now integrated in a way that is RESPA compliant and add value to all three parties.

Sramana Mitra: Interesting. Can you talk about your data sources for both automotive and real estate? What are the data sources that you are drawing from?

John Price: In automotive, it’s everything. We get automotive data everywhere. We get it from all the feeds that are out there. We get them directly from dealers and websites. Data is everywhere in automotive, but it’s all formatted differently.

In automotive, they call the data that the dealer types in as seller’s comments. We parse out all the valuable features from that and add those to our dataset. As a result of all those sources, we may see a car in five different places. With our normalization capabilities and our image decoding, we’re able to create one of the best highest quality sets of data for that particular image. We might take in 20 million versions of cars a day that represent 4 million unique cars in the market. We throw it on our data vault.

As a result, I have a 10-year history. As a result of that, I can drive a lot of correlation. The second part of my automotive data is, I have all the demand data. We saw all this traffic for all major brands. We were running Yahoo autos. We have 300 automotive consumer websites including our own.

Sramana Mitra: Earlier, you said if it’s an ad-supported model. The problem you have on the demand data side is if the data you’re receiving is influenced by ads on the site, you’re not getting a clear signal of consumer shopping behavior.

John Price: We always use what inventory is most likely to convert in this particular presentation of search result. As a result of that, we have unbiased, clear signals coming from tens of millions of shoppers across the web and mobile. As a result of that, we’re able to have extraordinarily accurate demand analytics over the inventory. They’re all up in Hadoop. That’s where the automotive data is coming from today.

When you go over to real estate, you’re working with the LMSs of the country. They have their own idiosyncrasies. You have to work with them very closely and keep up with the data on a daily basis. We keep around 4 to 5 million units that are for sale. Everything else being the same, the beauty of what we’re doing is it’s one stack and two very similar verticals.

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