 Okay, we're back here live in San Francisco, California's Oracle Open World 2012. It is packed, the streets are closed, this is SiliconANGLE.com's theCUBE, our flagship telecast where we go out to the events and extract a signal from the noise. And boy, there's a lot of noise here at Oracle Open World. They've got the streets are closed, 50,000 people, Oracle's waving their hands, jumping up and down. Cloud is here, big data's here. I'm John Furrier, the founder of SiliconANGLE.com, excited to be here. I'm joined by my co-host. I'm Dave Vellante at Wikibon.org. We're here with Sean Belvins, who is the executive vice president and general manager of Opera Solutions, a big data company. We found out about Opera when we did our big data study. Jeff Kelly quantified the marketplace and he did a little section on pure plays and Opera was one of the top two pure play, big data companies. Sean, we met at SAP Sapphire. Welcome back to theCUBE. I'm glad to be here. Thank you for the tremendous lift we got out of explaining the value and impact that we bring at Sapphire. How are you guys, a special debt? All right, well, we were excited because when we met you, we were like, okay, new person to theCUBE, and theCUBE's great. Extract the ceiling from the noise, get the big ideas out there, but you really had a value proposition that was at the beginning of, not the beginning of, in midstream of the big data revolution. Obviously SAP had huge success with HANA, that messaging, and I don't think they took advantage of the big data positioning as much. It kind of took more mobility, but you were talking about stuff that was so relevant. Now looking back, it feels like a decade since SAP, and the market has completely tipped over and completely validated around big data. So a couple of things. One is you have news here. You've worked with SAP, that's where we met you. Tell us what's happening here at Oracle. Then we want to go into some of the landscape changes in the marketplace. Absolutely, so first and foremost, we are announcing here at this event a partnership with Oracle around their big data appliance and exolytic stack. So they have two core capabilities that we like to say makes the conversation of the machine flow. So the essence of what we do is not analytics, it's taking all of the sources of big data inside and outside the four walls of your business, applying them in a signal hub, which is a collection of science models from our 230 scientists team in a productized, delivered as a subscription service way. So we're not backing a bus up, we're not bringing a bunch of consultants. This machine takes that big data, runs the science models against it and produces a set of best actions. It literally speaks and says if you do this, you can raise your top line or bottom line profit. Here's where the savings are, here's the actions to take with your customers, here's how you should price, here's the fraud and risk in your business. And so what happens with the Oracle announcement is that conversation between the machine and the person speeds up. It doesn't become a stuttering machine, it becomes a flowing machine where you can get real-time complex event processing. You don't have to wait for the machine to do all the scraping across the web of all these sources and then run those models. We can actually stage the models right inside of the exolytics box and then the big data appliance box with Hadoop brings all of those sources and loads them in lightning fast speeds. So it just makes for the machine speaking much faster and we're doing some work together in their solution center to really validate that to Oracle customers. Okay, so basically the signal layer is what you're laying on top of exolytics. And that's really most of the big data IP that you're bringing, right? I mean, the exolytics platform, other than I guess package with Hadoop, but essentially it's a block of infrastructure that you guys are leveraging. You got it, that's exactly right. And one of the nicest analogies I've heard is, if you go to play a video game, you need two things. You need the Xbox or the PlayStation and it's really just a box of chips and really fast IO and graphics, but you don't have the game. So Opera's released a signal hub suite, which is not only that layer, but a set of applications that we can talk about that are those video games on top of that console box. And the difference I think with Oracle is that they didn't take the SAP route, they didn't re-architect. So the stuff that's in their structure plays out of the box with the heritage of the other architecture that you find. So customers can deploy the big data stack from Oracle and then use our signal hubs immediately on top of that. John, what's your perspective on the legacy companies, the dinosaurs some people call them, SAP, Oracle. You don't necessarily associate those guys with big data. You do associate Opera with big data and obviously Cloud Air and Hortonworks and companies like that. What's the role of these large companies? Is it you work with them because they got customers? I mean that's one obvious choice, where are they in the whole big data landscape in your view and why do you sort of gravitate toward these big whales? Well we're working across the ecosystem, we're essentially agnostic in that we work with whatever we find with the customer because big data is data that you can't control, it's unstructured, it's outside your four walls, you don't own it, right? So what we're finding is that inside the four walls there's a triangulation of the data they have across all of the purpose build applications in ERP and CRM and things like that. So they have the, if you will, the keys or some of the metadata that we would say, okay we want to know who your customer is but they can't, they don't have the capability to see what that customer's doing 23.9 hours of the day, they're not engaging with this company. So what are they buying, what are they looking at, what other things are they wanting to do? So for a very large airline for example, we use this technology and now when you come to their site, we know that you've been looking at Bermuda and you rent a Hertz car and you stay at the Marriott and you like French food and you like to play golf so it makes you an offer on this airline based on what you've actually been looking at before you came to their site tied with the data that they have. So we take the heritage of the data in these systems and marry it to what we consider to be the other 95% of data that they simply don't have any capability to even get to. So Signalhub's, the best way to say it, is Signalhub brings the 95% of data, you don't have any capability to leverage and mix it with the 5% that you have spent hundreds of millions of dollars to do transactional process. And that data source comes from where? Is that something that you recommend? Is that something that the customer brings to the table? That's a heck of a point. The reality is in this new world, the application is not a database schema and it's not Java code, the data is the application. So I'm looking for fading and attrition or I'm looking for elasticity and pricing. The data informs the application and the science models, the secret of our 230 scientists. Google's got 300, we've got 230. All of ours are pointed in different domain in industry and commercial space. And we take those models and when someone says this is where I want that missile to land, we go from an outcome base and then we say okay, we need these 50 terabyte or larger sources, Facebook, Twitter, social, unstructured, internal, wherever it comes from, runs that models through and that's what informs the best action. And that best action is when the machine doesn't come back with a bunch of numbers, it comes back and says pick up the phone, call Sally Smith because she's in danger of canceling that subscription online. So without naming names, some very large weight management places, dating services, we can tell you when these people are fading and they're getting ready to cancel a subscription so that you can take action. And the way we do that has to do with understanding behavior, sentiment, time series, clustering, segmentation. All these models have to get stacked in such a way because if you don't, the other approach is that you buy a tool from a vendor like SAS and you buy a skill set and he works it and you get a single digit two or 3% lift from statistical analysis. We're not talking about that. We're talking about a model where the machine, it writes a script and says, when you call Sally Smith, here are the five things you need to say to her and here's the offer you need to make to her. So much more measurable productivity in fact. Oh, completely, completely. Sean, so you know, most people out there that either in business, they get an iPad, they go, wow, I see the real time web and we talked about this at Sapphire where you're traveling out real time mobility. They see the Siri on the iPhone. Wow, magic. That's magic's happening. So that brings up the whole low latency. We all know that. We've been talking about this slash, et cetera. And Larry's talking about here at the keynote here about the performance and SAP same thing. It's all about performance in memory. The machines are getting faster and faster and faster. And that's great. That's good for business, good for everyone. However, we're data jockeys like you. We understand data. And when you have the ability to scale infinitely or somewhat infinitely, bad data scales just as good as good data. So explain to the folks out there this concept of why this ontology or why having some experts in the verticals matter because our philosophy is big data is all about low latency value. And that it's not just storing data because if you don't clean the data, bad data scales just as fast as good data. So when you have a lot of bad data, you're spending more times figuring out how to make it good if you just do your homework on the front end and get those seeds. Because that's really kind of what you guys are doing. We call it signal, right? And a signal is, you know, first of all, you don't own the data. You can't control the structure. You can't store enough of it. You have to derive a signal, which is a pattern of behavior relevant to the outcome you're looking for. So signals are informed by the use case or the scenario. So when we do for a very large retailer, I'll give you a perfect example. This retailer is up against the Walmart's and the targets of the world, but they're especially retailer, right? And they have to price a skew and that skew can't be priced higher than a target or a Walmart, right? So they have to get that pricing exactly right to be competitive. What we're finding is in real time, you can go into these other retailers and you can see for a geographic location when that skew is now sold out. So I go to that site and I say, I can't get Tickleby Elbow anymore because it's not available in this area. So at that point, the other store can raise their margin because they're the only one that has the inventory for that period of time. So they have to lower their price when it's in stock locally and then they can raise it when they're the only ones that have it, right? And that's the kind of situational fluency that you can't build an application fast enough for. You have to look at the data models and you're finding signals like they're out everywhere in the 10 mile radius. We're the only one that has it so we can actually get more margin for this. And then conversely, we can't be elastic here because it's across the street for $20. How do you compare and contrast to say Palantir, which has been getting a lot of buzz. They're coming in saying we can architect this big data operating system for application development. Is it similar to that, different? Very different and here's why. There's, as I said, the Palantir model and the other models are essentially consulting models. They have not industrialized and productized this offering in a way that allows us to do this in weeks instead of months. So they would do a project and they would say, we'll give you a 15% lift. And the problem with that is it costs you 8% to 10% to get the 15% lift and the time to benefit is months. We have got these signal libraries built in the Signal Hub for these Signal Suite applications for the modules that we have so that you're talking about an eight to 12 week proposition. The case I just gave you with the airline was 55 days start to finish. They pay us a flat subscription fee just like Salesforce or any other SaaS app. So they pay a subscription in for a cloud based front end that spits out those best actions. So it's a completely different software driven model that's productized versus a lot of discovery and build from scratch. And the secret sauce behind this is the platform where all of the data scientists have put those models in there so that they derive from each other. I would go back to, if you want to see a real proof of this, go back and look at the contest like Netflix where we came in first and second across 40,000 teams. We're doing some other contests, you may have heard of like KDD where it's all about big data models and getting these results. The one I find most interesting that proves our value is heritage health, was you have a patient who's a heart patient, they admit themselves to the hospital, Medicare pays for it, but if you discharge that patient in the next six months and they come back, they don't pay a dime. So you have to have some way of determining when to discharge, when to keep. And that's really what separates us is we do this in a software model with fixed cost, fast time to return versus the smarter planet which we like to think of as just a smart way to sell consulting. Tell us about the validation you've got. Obviously the marketplace is all gaga over big data and it is about value proposition. It's about industrializing the solution versus some academic or consultative approach. So check the box, just give us a quick update, funding. You guys have a huge amount of funding. How big is the company? Just give the quick buy the numbers, just run down the list. Well, we are private, but we did receive $84 million in funding from Silverlake KKR and so those two companies continue to drive a lot of our growth. We're having tremendous growth across private equity portfolio companies because when you buy these companies, you have a couple of big issues. When you buy a portfolio companies and bring them into your group, you want to drive, you want to get savings. So we have a solution called spend intelligent that looks at every line item and every invoice and every SKU and every contract temp labor in real time and says, while you were out, I found $8 million on these SKUs, bought off contract or you didn't get your discount or here's temp labor where you're getting billed, not according to your negotiated rates. So when these private equity companies bring us in, they use this immediately for bottom line savings and then for top line growth. So there's a symbiotic relationship there and we're seeing another round of financing come through where people are wanting to invest and we're basically trying to negotiate that now. Sales this year and next year are over 40% growth. That's being conservative but we're private so I honestly don't know how much I'm allowed to say. The growth has been exponential. Give some color on the traction. I mean, I don't want to reveal your secrets, but yeah. The biggest customers are we have about, we've announced Morgan Stanley at Sapphire but some of the other ones, if you go on the site and you read Schwann's, AmEx, American Express, all of the ones that you would suspect for fraud and risk, all of the credit card companies use us for underwriting, fraud, risk, something called bust out, which is very interesting. You can see instead of someone stealing an identity, they create an identity out of whole cloth. They get a card, they pay the card, they continue to pay the card, they get a bigger credit line, now they have $100,000 in credit, then all of a sudden it flips and they go to jewelry stores and Best Buy and start getting stuff for this identity that doesn't exist. So there's really, unless you're able to go look and see that this person really doesn't exist in the real world and that their behavior and their payment doesn't match the behavior and payment of a segmented person like them, you wouldn't be able to catch that. So in our world, we take all these data sources externally and we say, okay, based on how they pay and what they paid and the amounts they paid over time and then what they didn't buy before they got more credit, this person is getting ready to bust out and go buy a bunch of stuff. So we can cut those cards off prior to that purchase. So that's an example. I want to contrast and compare that to a traditional consulting model. You talked about, you took a shot at IBM there. I did, didn't I? And so yeah, you did. I want to explore that a little bit. So a classic consulting model is what's your problem? We can solve it. We can find whatever data you need, we can do whatever you need to do, we can develop whatever application you want. Is your ability to respond more quickly, largely a function of your focusing on specific domain areas on industries and you've got data sources built up in those industries and those models are very industry specific or is there something else? Yeah, I would say the two things. First of all, we have, from the beginning, hired data scientists and given them freedom, said, listen, we don't care what models and technologies and tools you use, but you have to prove to this community of scientists that we have here that you are bringing additional value we don't have. So we don't have level two or level three scientists. We have the MIT's, the Harvard's, the Oxford's, the people that literally are the best of the best and we have 230 of them. So all of them work in a common platform and these models are deriving. So the difference between us is the stacked ensemble and then the ability to rapidly define and extract signal from all of these sources. The other guys, they don't necessarily talk about signal because signal is something that you have to have data science to deal with. So they want to build a one-off project and in most of the cases, they want to bill you as a consultant for that project. They're not giving the customer the ability to have a signal hub so that when those signals come in, it becomes an app store, right? Just like an iPad, you know, I'm reminded of Steve Jobs. I watched an interview with him a while back and he said, someone asked him, they were making a TV show about the future and I said, what is the real future gonna be like? And he said, well, the big shift is, and this is 20 years before anything, iPad or iPhone or anything, he says, you can't hierarchically move things around to adapt to change. I can't make someone else your boss fast enough. I can't change the organization and I can't change geographically. I can't move people around. He says, but once I know the outcome I want, it's very, I can very rapidly assemble the data and then for a specific purpose and pull people and resources around that data, get the answer and then blow it up and do it every 15 minutes. And what we've created with Signal Hub is the ability to say, for this outcome, you want to put this missile on this house 3,000 miles away, what are the sensors that we need to keep that missile on the right trajectory? What are the input sources that they need to hit that goal? We go outside the technology directly to the C-suite and we say, Mr. CMO, you want to grow sales 15% or you want to grow your market share, you want to take share, and we just devoid the whole technology and consulting model and do it in a very straight way that says these raw data sources plus our science models generate Signal which informs us of an action that the customer is likely to take and then we score it, we build millions of models and simulations, we kill them all except for the top 2% and then we come back to the customer and we say there's a 98% chance if you take this action, you will get that result. So that's very different, it's machine-based. But it's very much a consultative sell, right? How much, how much, yeah, so it's repeatable. How much demand for customization do you get and how do you accommodate that? Is the machine doing that customization? And or are you sort of just picking your spots and finding that repeatable model, in like customer segments? You're asking a great question, so in the case of the signals themselves and the Signal Hub suite, we have a suite that says, okay, here's customer action, so here's a 360 degree real-time view of your competitors, here's price, dynamic, elasticity, here's individualized, personalized offers, right? And those are very repeatable solutions that 90% of the people we talk to say I want to know in real-time, all these things about my customers, I want to know how to price my products, I want to make the right offers, and I want to take market share. So I was with a company about two months ago, and to your point, they use IBM, they use consulting, they use all these different approaches, they use lots of SaaS technology. And the guy, we went through this whole process and the guy says, we have everything. And they were a very big company, big hotel chain, and we do that. I said, you don't do that. He says, why don't we do that? And I said, well, it's really simple. You have campaign software, you have CRM software, customer marketing software, data warehouse, right? And I said, but I stay 200 nights a year at a competing hotel chain. You don't have me in your systems. I'm not one of your people, I'm not part of your campaigns, I'm not part of your loyalty system. But when I post on Twitter and Facebook that I'm mad that I had to pay for wifi and I didn't get a free breakfast, and I had to pay for parking, that's a signal. That's a moment you need to come to me and say, I'll honor the tier of service you're on, I'll give you free wifi, come over here and stay with us, right? So the data is not what we're talking about. And this is where I think the market misses it. The idea of signal and a hub that can bring those in and identify those moments of truth, that moment where that signal turns into an action take, that's the industrialized product we put together. So for each of these, whether it be savings and sourcing or operational spend, I love this model, we did this where a credit card issue where we looked at how they're paid and how their customers paid it. And we noticed that some of these people were effectively OCD. They just the moment they get their bill they pay it, right? So we're talking about millions and millions of dollars. So by simply segmenting out and figuring out sentiment, time history, how they paid over time, we said if you bill these customers first, you will get severe float on that money for 30 days. So take the people we've identified with that behavior, put them at the front of the line, bill them now and now you've got 25 days of float, about $400 million. So it's a business thing and what I want to bring your attention to is we did that in 30 to 40 days. So some of that is sort of more traditional but more powerful sort of analytics on real transactional data. The other piece that you mentioned is almost inferring from gesture data in real time defined as before you lose a customer or before you can't gain a customer. How do you price, how do you charge for your services? I would, the first thing you need to know is that if you're buying enterprise software, it's exactly the same. You're going to buy a license for this hub where the models are and the ability to bring in the signal. So we have something called Intelligent ETL where all your internal purpose build apps and all of the signal from sources you don't own outside the four walls, they're brought together into the hub so you license the hub and then you simply pay us a fixed fee as a subscription much as you would work day or SaaS Salesforce. And the difference between us is we're accountable. I mean, we'll take game share and I'm in three negotiations right now with customers who did that. And then- Oh, they're probably sorry. And they're very sorry. We didn't expect you to bring us 30, 40% of return. We thought you're going to give us 13 or 15, right? So- So you love those deals. I love those deals. So if you're out there today and you don't believe this, call me. Let's turn, you know, let, once the machine returns. You'll G-risk that transaction. I'll be happy to do that. So let's talk about some of the realities of the marketplace. I'm just looking at our little Twitter dashboard here as a tweet up here that says, big data without defining success first is a big mistake. I love that. That brings us up to the point where, you know, we're in similar businesses where, I mean, not similar business, but similar early adopter, you know, groundbreaking models. And there's always the naysayers, right? So I want to ask you specifically, you've had success in this area. How do you explain to a prospect and a client, discretion, yeah, we want to do this, but they're skeptical. Prove it to me. So the old way was, you mentioned Steve Jobs, you know, if I know my outcome, I can configure for it. Right. The old way is, you know, I need evidence. Can you show me evidence from another client or are the proof? Kind of like pushing a rock uphill. Right. What do you say to those people? How do you sell into those environments where you know they need your solution? They kind of like not accepting it. Is there a way you talk about it? Is there a way that? I actually, that's a hell of a point. I think the reality is that, you know, to some people, what we're talking about is UFOs. This is time travel. This is teleportation, right? You walk into a company that's spent a half a billion dollars on technology and you say you've got, you're effectively asking one question, which is, is there enough milk in the refrigerator? Do we need more milk? And you're asking the wrong questions because all that technology gives you is three to five percent of the total available data, right? So it is hard for people to understand really what they can do. I mean, in the government, we've got threat assessment that can see the difference between a bad actor and someone visiting a sensitive jihadi site and make assessments based on person's past behavior if they are a threat and be proactive with those people, right? We've got solutions that are just literally, you know, they're making up TV shows about us, right? We do this in real life. And so my result is twofold. First of all, we productize this and we have references. So you can pick up and speak with someone that is in the CEO, CFO, chairman role at these companies because the value chain with us is such that once this happens, it's like planning a foundation in the business where they may not trust the technology, maybe they do it on Gainshare the first time and they let's put this out. Is it a POC or do you come in and engage them? Yeah, we've done some POCs and pilots. What we're trying to do now is literally we have the demos. It's a turnkey situation in eight to 12 weeks where, you know, we're bringing repeatability to this and what happens is the hub and the models and the signals, the only thing that really changes is the configuration of the data science models and the kinds of feeds that produce the signals. So once you have the hub and you've licensed the hub, what we're finding is across the suite, people say, okay, I want you to manage my liquidity. I want you to manage my spin and marketing and then I want you to look at my customer actions and then I'm also interested in you managing invoice fraud and risk. So they look across the business and they go, where else can we get exponential, not incremental, exponential returns? And because the data is unlimited, we can keep adding those sources until we find enough to make the puzzle complete and say here is a 20 or 30% return. So what we're effectively doing for someone that's skeptical is we will do a pilot with them, but we'd much rather have them speak to a customer who is in a negotiation to move from gain share to fixed fee because they're not going to pay us anymore than what they've been paying us. It's an amazing technology that really empowers an executive to bypass the noise, right? I don't care about what app to buy or what piece of software to buy. I want to sit down with you, Opera, and say here's where I want strategically to grow my top line. I want to take share, I want to make sure and drive behavior at a pet store. We drove staples, so everybody that owns a cat has to buy cat food and cat litter. Very basic thing, but they got to buy it every so often. So by frequency, we looked at that and we said, when do your competitors and how many different pet stores and how many different people that sell that, do you have to drive by between this person that you're marketing to, house, and your pet store? And then when do they do their offers and how do we preempt that and when is their frequency of buying so that we drove them to buy, and we twinned them up with someone else who does buy all those staples from you so then we could very rapidly, and that's, but with us, it's a built solution, right? It's not something we have to build. So you get time to benefit in eight weeks to 12 weeks. Sean, we're getting up on the time limit here, but I want to ask you a few final questions. One is, we've met you at SAP, you saw that whole show, you have a business relationship with SAP, now you're announcing a relationship with Oracle. This is actually what Oracle wants, they want to have a lot of applications, they want us to be the Steve Jobs of the enterprise, he needs apps to come in. So with that, I want to ask you a question. For the folks out there that aren't really following the kind of the inside the ropes, what's going on between SAP and Oracle? Share with them the perspective of both shows. SAP Sapphire, their show, and this show. You want the skinny, you want the scope. Yeah, we'll see how you handle this one. Oh my. Share with them just a perspective, just some color, not necessarily, you don't have to throw the dagger, you can just say, SAP's got this, Oracle's got this, different philosophies, just what's your perspective? Okay, so without putting words in my mouth, because that's not sanitary, the reality is that both of these companies, in my mind, have recognized that they have to be in the space of big data. The problem is that big data in our assessment, having worked with this for eight years as a company, 700 people, 230 data scientists is you, big data is essentially just data that you can't control. The moment that you have a methodology and an approach for harnessing information raw data, which 90% is noise, 95% is noise, maybe more than that. So where I think these companies are evolving to is they're moving up the value chain to say, okay, it speeds and feeds and messing with the data, and this whole notion of storage and the whole notion of being able to capture things, it's really about, I think the evolution where the two companies are, we've got to find a way to put a canoe in the stream of data and put the sensors in the water and get the results we want. And that's the result of, when they look at our partnership, it really is opera has these things, turnkey, plug and play on our box. They got performance. Yeah, they got performance, they can go faster, they're building out an algorithmic layer so our models can live in the same box as the data side by side in real time and flash memory. So now, for K-Means clustering, we can put that right in the box, we can a-dupe load through big BDA, big data appliance from Oracle. So I think what you're going to see with these guys over time is they're going to recognize that data science is how applications come together. When you have an iPad, you want what you want, you don't want to have to learn a model, you don't have to learn a feature function or a UI. You want a very specific set of data presented to you for a specific action instead of outcome. And so the ability to put that algorithmic layer with these two vendors, that's where they're both, I think, in a race to right now. And the difference in my mind is that SAP has said to get there, we have to blow up the database paradigm. We have to do this in memory in a new model called HANA. And I think Oracle has said we have a very vast customer base and we want to bring them along with our existing model. So I think what happens is they're going to meet in the middle because at some point very soon, HANA will run underneath SAP. And at the same time, HANA will be available for new application development. They'll be there, but right now, if you are a current Oracle database customer, you can plug and play big data appliance, plug and play exolytics, and those other things, and then plug signal hubs on top of it. So I think from a plug and play disruption perspective, you could be as an Oracle customer, it doesn't require as much of a re-architecture, but from an SAP customer, you will have a green field for big data in a bigger way. So it's a trade-off in my view, and it comes down to flavor. I think both philosophies are right. I think they're going to get where they want to be. What we're interested in is, are you able to stage those data science models? And are you able to give us a simple environment so that we can make those conversations with the machine flowing and not stuttering? It's a good answer. You're an arms dealer for both SAP and Oracle. I love them both. I worked in both for seven years. Seven years in SAP and six years at Oracle. You win either way. It's good for your business. It's good for the value proposition that you're offering. We love your business. Dave and I, two conversations, we think you guys are on a path that everyone should look at and say, hey, that's a path of the future. Big data, using data, connecting the applications. Sean, thanks for coming on theCUBE again. CUBE alumni. The data is the app. You heard it here. The data is the app. Certainly a tech athlete. The kind of guests we love on SiliconANGLE.com and Wikibon. We'll be right back with our next guest, CEO of DataStacks, right after this short break. Thank you.