 Hey, welcome back everybody, Jeff Frick here with theCUBE. We're having a CUBE conversation in the Palo Alto studio. It's a different kind of format of CUBE, not in the context of a big show. Got a great guest here, lined up who we just had on at a show recently. He's Nathan Trueblood. He's the Vice President of Product Management for Data Torrent, Nathan, great to see you. Thanks for having me. We just had you on theCUBE at Hadoop, or DataWorks now, not Hadoop Summit anymore. So just quick follow up on that. We were just talking before we turned the cameras on. He said that was a pretty good show for you guys. Yeah, it was a really great show. And in fact, as a software company, one of the things you really want to see at shows is a lot of customer flow and a lot of good customer discussions. And that's definitely what happened at DataWorks. It was also really good validation for us that everyone was coming and talking to us about what can you do from a real-time analytics perspective. So that was also a good strong signal that we're onto something in this marketplace. Yeah, it's interesting how this, heard from someone that really the streaming and the real-time streaming in the big data space is really grabbing all the attention. Well, obviously we do Spark Summit. We did Flink Forward. So we're seeing more and more activity around streaming and it's so logical that now that we have the compute horsepower, the storage horsepower, the networking horsepower to enable something that we really couldn't do very effectively before, but now it's opening up a whole different way to look at data. Yeah, it really is. And I think as someone who's been working in the tech world for a while, I'm always looking for sort of simplifying ways to explain what this means, because people say streaming and real-time and all of that stuff. For us what it really comes down to is the faster I can make decisions or the closer to when something happens I can make a decision, that gives me competitive advantage. And so if you look at the whole big data evolution, it's always been towards how quickly can we analyze this data so that we can respond to what it's telling us. And in many ways that means being more responsive to my customer. So a lot of this came out of course originally from very large scale systems at some of the big internet companies like Yahoo where Hadoop was born. But really it all comes down to if I'm more responsive to my customer, I'm more competitive and I win. And I think what a lot of customers are saying across many different verticals is real-time means more responsiveness and that means competitive advantage. Right, and even here all the time, moving into like a predictive model and then even to a prescriptive model where you know you're offloading a lot of kind of the grunt work of the decision making, letting the machine do a lot more of that. And so really it's the higher value stuff that finally gets to the human at the end of the interaction who's got to make a judgment. That's exactly right, that's right. And so to me all the buzz about streaming is really representative of just this is now the next evolution of where big data architecture has been going which is towards moving away from a batch-oriented world into something where we're making decisions as close to the time of data creation as possible. Right, right, so you've been involved not only in tech for a long time but Hadoop specifically and Big Data specifically. Yeah. And one of the knocks, I remember the first time I ever heard about Hadoop is actually from Bill Schmarz, ODMC, the Dean of Big Data and I was talking to a friend of it, he goes, yeah, but I would build it and tell you there's nobody, there's not enough people. You know Hadoop's got all this great promise, there just aren't enough people for all the enterprises at the individual company level to implement this stuff. A huge part of the problem. And now you're at DataTorrent and as we talked before interesting kind of shift in strategy and going to really an application-focused strategy as opposed to more of a platform-focused strategy so that you can help people at companies solve problems faster. That's right and we've definitely focused especially recently on more of an application strategy but to kind of peel that back a little bit. You know, you need a platform with all the capabilities that a platform has to be able to deliver large scale operable streaming analytics. But customers aren't looking for platforms, they're looking for please solve my business problem, give me that competitive advantage. And I think, you know, it's a long-standing problem in technology and particularly in Big Data where you build a tremendous platform but there's only a handful of people who know how to actually construct the applications to deliver that value. And I think increasingly in Big Data but also across all of tech, customers are looking for outcomes now and the way for us to deliver outcomes is to deliver applications that run on our platform. So we've built a tremendous platform and now we are working with customers and delivering applications for that platform so that it takes a lot of the complexity out of the equation for them. And we kind of think of it like if in the past it required sort of an architect level person in order to construct an application on our platform. Now we're gearing towards a much larger segment of developers in the enterprise who are tremendously capable but don't have that deep Big Data experience that they need to build an application from scratch. And it's pretty interesting too because another theme we see over and over and over and over, especially around the innovation theme is the democratization of the access to the data, the democratization of the tools to access the data so that anyone in the company or a much greater set of individuals inside the company have the opportunity to have a hypothesis, to explore the hypothesis, to come back with solutions. And so by kind of removing this ivory tower, either the data scientist or the super smart engineer who's the only one that has the capability to play with the data and the tools, you know that's really how you open up innovation is democratizing access and ability to test and try things. That's right, to me I look at it very simply when you have large scale adoption of a technology usually comes down to simplifying abstractions of one kind or another. And the big simplifying abstraction really of Big Data is providing the ability to break up a huge amount of data and make some sense of it using of course large scale distributed computing. The abstraction we're delivering at Data Torrent now is building on all that stuff on all those layers. We've obscured all of that and now you can download with our software an application that produces an outcome. So for example, one of the applications we're shipping shortly is a omnichannel credit card fraud prevention application. Now, our customers in the past have already constructed applications like this on our platform. But now what we're doing like you said is democratizing access to those kinds of applications by providing an application that works out of the box. And that's a simplifying abstraction. Now truthfully, there's still a lot of complexity in there but we are providing the pattern, the foundational application that then the customer can focus on customizing to their particular situation, their integrations, their fraud rules and so forth. And so that just means getting you closer to that outcome much more quickly. Watching your video from DataWorks, one of the interesting topics you brought up is really speed and how faster, better, cheaper, which is innovative for a little while becomes the new norm. And as soon as you reset the bar on speed, then they just want it to go faster. So whether you went from a week to a day, a day to an hour, there's just this relentless pressure to be able to do, get the data, analyze the data, make a decision faster and faster and faster. And you've seen this just changing by leap years, right? Right, I mean I literally started my career in the days of ETL extracting data from tape that was data produced weeks or months ago down to now we're analyzing data at volumes that were inconceivable and producing insight in less than a second, which is kind of mind-boggling. And I think the interesting thing that's happening like when we think about speed, and I've had a few discussions with other folks about this, they say, well, speed really only matters for some very esoteric applications is one of the things that people bring up. But no one has ever said, well, I wished my data was less fresh or my insight was not as current. And so when you start to look at the kinds of customers that want to bring real-time data processing and analytics, it turns out that nearly every vertical that we look at has a whole host of applications where if you could bring real-time analytics, you can be more responsive to what your customer is doing. Right, right. And that can be, certainly that's the case in retail, but we see it in industrial automation and IoT. All I think of as IoT is a way to sense what's going on in the world, bring that data in, get insight and take action from it. And so real-time analytics is a huge part of that, which again, healthcare, insurance, banking, all these different places have use cases. And so what we're aiming to do at DataTorrent is make it easy for the businesses in those different verticals to really get the outcome they're looking for, not produce a platform and say, imagine what you could do, but produce an application that actually delivers on a particular problem they have. It's funny, too, the speed equation, and you saw it in flash, memory to shift gears a little bit into the hardware, so people said, well, it's only super low latency, super high volume transactions, financial services is the only benefit we're going to get flashed. Yeah, we've had the same knock for real-time analytics. Same thing, right? But as soon as you put it in, then there's all these second-order impacts, third-order impacts that nobody ever thought of, that speed that delivers, that aren't directly tied to that transactional speed, but now enable you, because of that transactional speed, to do so many other things that you couldn't even imagine to do. And so that's why I think we see this pervasiveness of flash, why wouldn't you want flash? I mean, why wouldn't you want to go faster? Because there's so much upside. Yeah, so again, all of these innovations in IT come down to, how can I be more flexible and more responsive to changing conditions, more responsive to my customer, more flexible when it comes to changing business conditions, and so forth. And so now as we start to instrument the world and have technologies like machine learning and artificial intelligence, that all needs to be fed by data that is delivered as quickly as possible and then it can be analyzed to make decisions in real time. So when it shift gears a little bit kind of back to the application strategy, so you said you had the first app that's going to be the fraud prevention. Yeah, so the first application were, yes, it was fraud prevention, and that's an important distinction there because the distinction between detection and prevention is the competitive advantage of real time, because what we deliver in data torrent is the ability to process massive amounts of data in very, very low timeframe, sub seconds time frames, and so that's the kind of fundamental capability you need in order to do something like respond to some kind of fraud event. And what we see in the market is that fraud is becoming a greater and greater problem. The market itself is expanding, but I think as we see fraud is also evolving in terms of the ways that it can take place across e-commerce and point of sale and so forth, and so merchants and processors and everyone in the whole spectrum of that market is facing a massive problem and an evolving problem. And so that's where we're focused in one of our first, I would say, vertically oriented business applications is it's really easy to be able to take in new sources of data with our application, but also to be able to process all that data and then run it through a decision engine to decide if something is fraudulent or not in a short period of time. So you need to be able to take in all that data to be able to make a good decision and you need to be able to decide quickly if it's going to matter. And you also need to be able to have a really strong model for making decisions so that you avoid things like false positives which are as big a problem as preventing fraud itself if you deliver a bad customer experience. And we've all had that experience as well, which is your card gets shut down for what you think is a legitimate activity. It's just so ironic that false positives are the biggest problem with credit card fraud. You would think we would be thankful for a false positive, but all you hear over and over and over is a false positive in the customer experience. It just shows that we're so good at it is the thing that really urges people. Well, and so if you think about that, having an application that allows you to make better decisions more quickly and prevent those false positives and take care of fraud is a huge competitive advantage for all the different players in that industry. And it's not just for the credit card companies, of course, it's for the whole spectrum of people from the merchant all the way to the bank that are trying to deal with this problem. And so that's why it's one of the applications that we think of as a key example where we see a lot of opportunity. And certainly people that are looking at credit card fraud have been thinking about this problem for a while, but there's a complexity like we were discussing earlier of finding the talent on being able to deliver these kinds of applications, finding the technology that can actually scale to the processing volume. And so by delivering omnichannel fraud prevention as a big data application, that just puts our customers so much closer to the outcome that they want and it makes it a lot easier to adopt. Right, so as you sit, shift gears a little bit as your VP product had and there's a huge wide world of opportunity in front of you, we've talked about IoT a little bit, obviously fraud, you've talked about omnichannel retail. How are you guys going to figure out where you want to go next, kind of how are you prioritizing the world and as you build out more of these applications, is it going to be vertically focused, horizontally focused? What are your kind of thoughts as you start down the application journey? So a few thoughts on that. Certainly one of the key indicators for me as a product manager when I look at where to go next and what applications we should build next, it comes down to what signal are the customers giving us. And so as we mentioned earlier, we built a platform for real-time analytics and decision-making and one of the things that we see is broad adoption across a lot of different verticals. So I mentioned sort of industrial IoT and financial services, fraud prevention and advertising technology and we have a company that we're working with in GPS geofencing so the possibilities are pretty interesting. But when it comes to prioritizing those different applications, we have to also look at what are the economics involved for the customer and for us. So certainly one of the reasons we chose fraud prevention is that the economics are pretty obvious for our customers. Some of these other things are going to take a little bit longer for the economics to show up when it comes to the applications. So you'll certainly see us focusing on vertically oriented business applications because again, the horizontals tend to be more like a platform and it's not close enough to delivering an outcome for a customer. But it's worth noting one of the things we see is that while we will deliver vertically oriented applications that oftentimes switching from one vertical app to another is really not a lot more than changing the kind of data we're analyzing and changing the decision engine. But that the fundamental idea of processing data in a pipeline at very high volume with fault tolerance and low latency, that remains the same in every case. So we see a lot of opportunity essentially as we solve an application in one vertical to reskin it into another. So like tweaking the dials, tweaking the UI, tweaking the data and the rules that you apply to that data. So if you think about on the channel fraud prevention, well it's not that big of a leap to look at healthcare fraud or to look at all the other kinds of fraud in different verticals that you might see. Right. Do you ever see that you'll potentially break out the algorithm, I forget what you're talking about, algorithms as a service? Or is that too much of a bit? Is it need to be a little bit more packaging? No, I mean I think there will be cases where we will have an algorithm out of the box that provides some basics for the decision support. But as we see a huge market springing up around AI and machine learning and machine scoring and all of that, there's a whole industry that's growing up around essentially we provide you the best way to deliver that algorithm or that decision engine that you train on your data and so forth. So that's certainly an area where we're looking from a partnership perspective where we already today partner with some of the AI vendors for what I would say is some custom applications that customers have deployed. But you'll see more of that in our applications coming up in the future. But as far as sort of algorithms as a service, I think that's already here in the form of being able to query against some kind of AI with a question, essentially a model, and then get an answer back. Right, well Nathan, exciting times and your big data journey continues. It certainly does, thanks a lot Jeff. Thanks, Nathan Trueblood from DataTorrent. I'm Jeff Frick, you're watching theCUBE. We'll see you next time. Thanks for watching.