Chris Reddick and Ron Halversen, both of Clarity Ventures, talk about AI and machine learning in an eCommerce marketplace platform.

Part 4 of an 8-part series (Return to Part 3)

RON: One of the things you brought up that I loved, and you breezed by real quickly, as a rocket scientist, you always bring up these really nice little nuggets of gold—you mentioned AI for eCommerce marketplaces, but we really didn't dwell on that. That's important. And I know right now we're doing tons of research, we're doing videos, we're doing development work to embed a more robust—I mean, we've already had machine learning, where I can go in and we can program rules and say, “Hey, this vendor came in and they searched and they looked at nine hand tools and one power tool, and then they went away. And the next time they came, we served them up a 10% coupon off hand tools.” Why did we do that? Because the dwell time and the number of products that they looked at, 90% was hand tools. So we went and wrote a rule and blah, blah, blah.  
 
That's the stuff you can start doing to help increase [sales in marketplace eCommerce]. But take it even further, when you can start making recommendations. Ron goes to look at a laptop. Down the right-hand side there's three different lists. One list is, “Here's the top three selling laptops right now that are trendy.” That might be one list. The next list below that might be, “Here are the top accessories you're going to need for this laptop. Here's your RAM upgrade, here's your three-year warranty, here's the case, here's the HDMI cable, here's the external monitors, here's the external hard drive for backup.”  

 

what is hipaa

RON: And then the third one at the bottom might come up and say, “Here are the top sellers in the computer category.” And so it might have some tablets, maybe the MacBook Air, maybe some optional things that are not quite the laptop. But these are the top selling products for people that looked at this, this is what they actually bought, right? All of those would be and could potentially be dynamically driven and dynamically built on the fly as the user is going through and behaving in a particular way in your marketplace. Every single one of those things should and most likely would drive additional conversions and additional sales. So those are the types of things that I think are important. Do you want to talk about AI in eCommerce since I brought it up? 
 
CHRIS: Yeah, I think the best way to—and I’ll keep it relatively short, because I can just go on a monologue about AI—but the bottom line here is that we will continue to provide content around AI. There is truly a renaissance within AI. And AI is basically like the Russian matryoshka dolls, the nesting dolls. You have AI as a big bucket, then you have machine learning, and then within that you have deep learning. And a lot of the renaissance that's occurred over the last several years in particular, is based around deep learning in customized eCommerce solutions

Now, to boil it down really simply, if you recall, back in high school or college or middle school, whenever you did the fitting of data points to a curve, that's basically what AI does. It does it on multiple vectors or dimensions because it has a lot of data, and you’re looking at a lot of data. So it's not just a 2D graph that we're fitting the data points on. We're fitting it on multiple dimensions, but it really just ends up being basic matrix multiplication. It's not very complicated. But it does, whenever you get at scale, require relatively decent amount of focused compute resources.  

 

what is hipaa

CHRIS: What we're doing is, we're basically teaching through form-fitting of the data to equations. We build this set of equations through weights that we can apply to these different—it's literally just form-fitting the data to an equation that models it. And we have, in most cases, thousands to tens of thousands of different coefficients that define the curve and the slope of those different form fits. And this ultimately is what makes it so powerful is that a lot of the research over the last several years has become very open-source focused, very focused around providing the data models and the underlying coefficients and weights and process that people have used to get these incredible results that they're getting. 
 
So now, that basically is a requirement for anyone who wants to be very competitive over the next several years with marketplace software. That's going to just be a baseline. So you just need to know that if you're wanting to continue to be relevant and competitive over the next three to five years, you're going to want to incorporate some of this renaissance that's occurred. 
 
Someone's going to do it within your space. So is it going to be you or is it going to be your competitor? And just as you described, Ron, these capabilities essentially include learning based on the data, what users are expecting and what they would see as associated recommended items. If they're browsing for this item, they need to see the other items that other folks purchased and they’re happy with. Whenever sellers are going through the process, they need to have recommendations for gamification to keep them engaged on your marketplace website.  

 

what is hipaa

CHRIS: There is a spectrum of about 20 different topics that we will definitely will go into in a separate video. I highly encourage you guys to like and subscribe to this video [channel]. If you're interested in seeing those more detailed videos, you can leave a comment as well and we'll respond to you and let you know once the more detailed I based in machine learning, deep learning videos come out. 
 
Anyways, suffice to say, the bottom line here is that because it is something that is open source and is rapidly developing, you're probably going to continue to see an acceleration of this adoption within online marketplaces. And if you're not investing into it, you're probably going to end up at some point within your space, dealing with competitors who are going to trump you if you don't invest in it yourself. 
 
RON: Yeah, that's true. We've definitely seen that acceleration and that's why I've seen you double down on our investment in the research and bringing that more into our platform, where we were at the machine learning. Now we're into the deep learning and AI research.  

Continue to Part 5 to learn more about selling subscriptions and non-tangible services.