>> Announcer: Live from Orlando Florida. It's the Cube. Covering Pentaho World 2017. Brought to you by Hitachi Vantara. >> Welcome back to the Cube's live coverage of Pentaho World brought to you by Hitachi Vantara. I'm your host, Rebecca Knight along with my co-host, Jim Kobielus. We are joined by Geo Thomas. He is the director of It at Benefits Science a healthcare insurance analytics company. Thanks so much for coming on the Cube, Geo. >> Thank-you, thanks for having me. >> So Benefits Science is a company launched out of MIT, tell our viewers a little bit more about the company. >> Okay, so Benefits Science is a healthcare data analytic company which co-founded by MIT (mumbles). Doctor (mumbles) and Doctor Stephen so far and we have one more partner. We do data analytics on the healthcare side and we work with employers and the brokers to analyze the data and give them dashboards and workbooks, and so that's what we mainly do. And we, yeah. >> So, as you said, you work with employers to save them healthcare dollars. Can you get into the nitty-gritty a little bit more. >> That's exactly right, so what we do is we empower employers to manage their employee benefits. Providing them the data analytic tools and other optimization tools, and we give them a very fine clear picture of how these plans are performing, and how they can optimize their plans in the near future by giving plan optimization tools and (mumbles) algorithms and things like that. >> You refer this as a manage service for your clients or do you provide specifically licensed software that helps them do this for themselves? From their own premises. >> We are a Cloud platform, and we provide our platform as a sub-lease for our clients. So, we get the data from them and we provide data analytic tool by mashing of this data and they use our platform to see those reports and insights and things like that. >> So, healthcare data is a really special kind of complicated when it comes to data because there's so many security and privacy issues related to it, how do you go about it managing this kind of data? >> Healthcare data is a very complex, very huge and we can't expect what comes next and there a lot of regulations and there are a lot of security issues, so we take all these with upmost priority. So, our company is a SOC1, SOC2, certified company. Which covers a lot of regulations by itself. Our employee's, Benefits Science employees, are really very much aware of these heap of rules. And they are all certified. We have lots of internal an external audits and regulations throughout the place so that would cover all this compliance issues, mainly. >> From an operational standpoint, how are you managing the day-to-day, day-in and day-out, do you provide a data warehouse within which you load it and then from which you do the analysis? What's the sense for how you architected your environment and then where how Pentaho plays into the overall picture? >> We take the data. Once we get the data, we measure the data. So, how we do those, we use Pentahos, and then two and two. Because it gives us a very standardized methodology to process this data, so we identify the PHP data. We sample it, scramble it, and then we do the (mumbles). And once the data element is done, and nobody touches any of those PA jobs or the jobs which we created with Pentaho, and we run this in a very secure environment in which we put all this transformed data into a data analytical platform. >> When you say scramble, you're referring to masking and anonmyzing the data? >> Correct, yes. >> That's what I assumed, you tell me, that's required by HIPA, that you do it that way? >> Yes, that's correct, yeah, yeah. So, we don't take all the data for the development. We take only the sample data, and then we scramble it and we (mumbles) all this information. >> So, what kind of results have you seen in your company since using Pentaho? >> So, I started in almost one year back and when we started, we had 20 tenants. Now, we have 200 tenants, so that's the summary of recently of what I'm seeing because Pentaho gives us lot of flexibility to standardize and make proper checks and balances throughout the data pipeline and we had created very huge test framework which can run automatically. So, all these things would benefit us to board a client because right now, onboarding a client would take less than a week. >> When you say test run automatically what sort of test are you referring to? >> So, we create test scripts, and we created a test suit framework by using Pentaho Jobs. And we schedule that. That test suit what we do is every, whenever any tenant comes in, developers can create N number of test cases and plug that in. So, it is growing and that will run automatically. Along with the PA jobs. So, that gives us a number of outputs and checks and balances and depending on the results we board the client. >> Saving healthcare dollars, spending healthcare dollars. This is really part of the national conversation. How much does Benefits Science really feel a responsibility to weigh-in on these issues. We heard a lot from the CEO this morning about how Pentaho really views its guiding principles as doing good in the world and bettering society. >> The double bottom line. >> Very true, very true, because as Benefits Science company our vision, our motto is not to just built some software and give to customers and get some money. Our vision is to help people or employers reduce the healthcare cost, so. Our data scientists built this great plan optimization tool or (mumbles) to provide employers to look at, "Okay, these "are the large claimant details, which means we might have "to go and find out the reasons and work with them "to reduce the cost." So, we are giving all the tools for them and another thing is the data (mumbles) analyzer our users love it, because we provided a simplified cube for them to drag and drop and create the reports and they can easily drag a couple of data elements and come up with, "Okay, these are the paid amounts "which we paid last month, and this has to go down." So, they can come up with their own strategies to make it down, at least, for the next year and on. >> In terms of user's being able to, in a self-service basis define their views and their reports. Do you take that intelligence that you gained from users and then bring that back into the basic service in terms of adjusting the data model? The set of canned reports or dashboards you provide? What do you do in that regard? >> Yeah, so we have a custom insight reports. Which will give pretty good idea about what this data meant to be for the customers. Like drag dashboards or large claimants or quality measures so things like that. We also have another data science group works on this AI tools or machine-learning algorithms to provide more predictive analysis. So, that would give users a different perspective of, "Okay, if we do this, we can reduce the cost." >> Is that WECA or? >> No, we are using. That's another thing I want to go back and tell them. There is a WECA here, we probably have to start using it. So, right now, we are not, right now we are using RN Python. There's something called (mumbles). So, that's what we use. >> What are some challenges that you are facing right now? What is keeping you up at night? What do you want the next versions of Pentaho to solve for you? >> I'm Director of IT, so I care about IT more than the business. So, my challenge is always how I can board more clients within a short span of time. The scalability, the security, how we can make it compliant. So, I was listening to that ATO, what are the new things coming in ATO? One of the main thing I was looking at is the scalability that is there is something called Worker Naught, that's got announced in ATO. Which you can scale as a docker, and you can spin off as many dockers as you want, and it will work by itself. That's fantastic, I'm really looking forward to get that scalability into our system. >> So, you're saying your IT environment. Your focused now more and more on a Cloud data environment that takes the application functionality and wraps it as containers? So, that's where you're going? And then you're saying that, I don't want to put words in your mouth, what you're doing is consistent with where Pentaho's going with their overall product platform? >> We are hosting an (mumbles) Cloud with Pentaho. So, Pentaho is also going into that direction. Makes me very happy because we are really looking forward to get that working in the Cloud. The thing is the. The Worker Naught, what they're talking about? Is what we were thinking of implementing on our own. So, now they have their own Worker Naught which we can just take and put it there. So, that's very good news. >> I wanted to ask you about the talent shortage in technology because that is something that the CEO talked about, Karen Perlich talked about, too. Is this real dearth of talent in data science. There was a piece in the New York Times just the other day that talked about how data scientists just a PHD can come out and make a half a million dollars in Silicon Valley. What do you think will be the real change and will get more and more graduates into this field. It seems as though the money should be enticement enough. >> That's a million dollar question though. We are in the same boat. >> You're a Massachusetts' based company, it should be. >> Even with that, we are finding a lot of difficulties to get some good data scientists. Because the moment you pass out as data scientist they're asking half a million, so. >> Literally I saw an article the other day. A good data scientist in Silicon Valley can fetch upwards of a half a million per year, so. Imagine in other regions, and now Massachusetts has no shortage of educated, smart people, but still. >> They have that level, then yes. These tools would help, and. Building that artificial intelligence on top of these tools would help, definitely, to have some sort of, not depending on data scientists so much. That even others can do those kind of things. >> So, you might not need the talent in a way. >> I'm looking forward to that because I was listening to your session in the morning. Very impressed with that because that's where I'm also trying to see where the world is heading to. >> So, you make recommendations to your clients about how they should start structure their healthcare insurance plans or employees. Do you have a capability right now within Benefits Science to basically embed a recommendation engine of that sort to help advisors on your staff to work with clients to recommend the right set of options or approaches pulling from the data, that's already there? >> Yes, that's already there. So, we provide recommendations for clients by using these algorithms. So, we have this plan optimization tool. Which will give you, if you do such and such things this is going to go down in the next year. Or there is a plan designed data. So, whenever an enrollment happens the main thing that they look at is what plan they have to sell at for their set of employees. So, every case is unique. So, we put a lot of historical data information and we put those machine-learning algorithms in there and then we come up with. We clean that model with all this data and we predict for each tenant. So, we have that right now. >> Geo, thanks so much for coming on the Cube. It's been really fun talking to you. >> Thanks for having me. >> I'm Rebecca Knight for Jim Kobielus. We will have more from the Cube's live coverage of Pentaho World, just after this. (calm electronica music)

>> Announcer: Live, from Orlando, Florida, it's The Cube. Covering PentahoWorld 2017. Brought to you by, Hitachi Vantara. >> Welcome back to Orlando, everybody. This is PentahoWorld, #pworld17 and this is The Cube, The leader in live tech coverage. My name is Dave Vellante and I'm here with my co-host, Jim Kobielus Donna Prlich is here, she's the Chief Product Officer of Pentaho and a many-time Cube guest. Great to see you again. >> Thanks for coming on. >> No problem, happy to be here. >> So, I'm thrilled that you guys decided to re-initiate this event. You took a year off, but we were here in 2015 and learned a lot about Pentaho and especially about your customers and how they're applying this, sort of, end-to-end data pipeline platform that you guys have developed over a decade plus, but it was right after the acquisition by Hitachi. Let's start there, how has that gone? So they brought you in, kind of left you alone for awhile, but what's going on, bring us up to date. >> Yeah, so it's funny because it was 2015, it was PentahoWorld, second one, and we were like, wow, we're part of this new company, which is great, so for the first year we were really just driving against our core. Big-Data Integration, analytics business, and capturing a lot of that early big-data market. Then, probably in the last six months, with the initiation of Hitachi Ventara which really is less about Pentaho being merged into a company, and I think Brian covered it in a keynote, we're going to become a brand new entity, which Hitachi Vantara is now a new company, focused around software. So, obviously, they acquired us for all that big-data orchestration and analytics capability and so now, as part of that bigger organization, we're really at the center of that in terms of moving from edge to outcome, as Brian talked about, and how we focus on data, digital transformation and then achieving the outcome. So that's where we're at right now, which is exciting. So now we're part of this bigger portfolio of products that we have access to in some ways. >> Jim: And I should point out that Dave called you The CPO of Pentaho, but in fact you're the CPO of Hitachi Vantara, is that correct? >> No, so I am not. I am the CPO for the Pentaho product line, so it's a good point, though, because Pentaho brand, the product brand, stays the same. Because obviously we have 1,800 customers and a whole bunch of them are all around here. So I cover that product line for Hitachi Vantara. >> David: And there's a diverse set of products in the portfolios >> Yes. >> So I'm actually not sure if it makes sense to have a Chief Products officer for Hitachi Vantara, right? Maybe for different divisions it makes sense, right? But I've got to ask you, before the acquisition, how much were you guys thinking about IOT and Industrial IOT? It must have been on your mind, at about 2015 it certainly was a discussion point and GE was pushing all this stuff out there with the ads and things like that, but, how much was Pentaho thinking about it and how has that accelerated since the acquisition? >> At that time in my role, I had product marketing I think I had just taken Product Management and what we were seeing was all of these customers that were starting to leverage machine-generated data and were were thinking, well, this is IOT. And I remember going to a couple of our friendly analyst folks and they were like, yeah, that's IOT, so it was interesting, it was right before we were acquired. So, we'd always focus on these blueprints of we've got to find the repeatable patterns, whether it's Customer 360 in big data and we said, well they're is some kind of emerging pattern here of people leveraging sensor data to get a 360 of something. Whether it's a customer or a ship at sea. So, we started looking at that and going, we should start going after this opportunity and, in fact, some of the customers we've had for a long time, like IMS, who spoke today all around the connected cars. They were one of the early ones and then in the last year we've probably seen more than 100% growth in customers, purely from a Pentaho perspective, leveraging Machine-generated data with some other type of data for context to see the outcome. So, we were seeing it then, and then when we were acquired it was kind of like, oh this is cool now we're part of this bigger company that's going after IOT. So, absolutely, we were looking at it and starting to see those early use cases. >> Jim: A decade or more ago, Pentaho, at that time, became very much a pioneer in open-source analytics, you incorporated WECA, the open-source code base for machine-learning, data mining of sorts. Into the core of you're platform, today, here, at the conference you've announced Pentaho 8.0, which from what I can see is an interesting release because it brings stronger integration with the way the open-source analytic stack has evolved, there's some Spark Streaming integration, there's some Kafaka, some Hadoop and so forth. Can you give us a sense of what are the main points of 8.0, the differentiators for that release, and how it relates to where Pentaho has been and where you're going as a product group within Hiatachi Vantara. >> So, starting with where we've been and where we're going, as you said, Anthony DeShazor, Head of Customer Success, said today, 13 years, on Friday, that Pentaho started with a bunch of guys who were like, hey, we can figure out this BI thing and solve all the data problems and deliver the analytics in an open-source environment. So that's absolutely where we came form. Obviously over the years with big data emerging, we focused heavily on the big data integration and delivering the analytics. So, with 8.0, it's a perfect spot for us to be in because we look at IOT and the amount of data that's being generated and then need to address streaming data, data that's moving faster. This is a great way for us to pull in a lot of the capabilities needed to go after those types of opportunities and solve those types of challenges. The first one is really all about how can we connect better to streaming data. And as you mentioned, it's Spark Streaming, it's connecting to Kafka streams, it's connecting to the Knox gateway, all things that are about streaming data and then in the scale-up, scale-out kind of, how do we better maximize the processing resources, we announced in 7.1, I think we talked to you guys about it, the Adaptive Execution Layers, the idea that you could choose execution engine you want based on the processing you need. So you can choose the PDI engine, you can choose Spark. Hopefully over time we're going to see other engines emerge. So we made that easier, we added Horton Work Support to that and then this concept of, so that's to scale up, but then when you think about the scale-out, sometimes you want to be able to distribute the processing across your nodes and maybe you run out of capacity in a Pentaho server, you can add nodes now and then you can kind-of get rid of that capacity. So this concept of worker-nodes, and to your point earlier about the Hitachi Portfolio, we use some of the services in the foundry layer that Hitachi's been building as a platform. >> David: As a low balancer, right? >> As part of that, yes. So we could leverage what they had done which if you think about Hitachi, they're really good at storage, and a lot of things Pentaho doesn't have experience in, and infrastructure. So we said, well why are we trying to do this, why don't we see what these guys are doing and we leverage that as part of the Pentaho platform. So that's the first time we brought some of their technology into the mix with the Pentaho platform and I think we're going to see more of that and then, lastly, around the visual data prep, so how can we keep building on that experience to make data prep faster and easier. >> So can I ask you a really Columbo question on that sort-of load-balancing capabilities that you just described. >> That's a nice looking trench coat you're wearing. >> (laughter) gimme a little cigar. So, is that the equivalent of a resource negotiator? Do I think of that as sort of your own yarn? >> Donna: I knew you were going to ask me about that (laughter) >> Is that unfair to position it that way? >> It's a little bit different, conceptually, right, it's going to help you to better manage resources, but, if you think about Mesos and some of the capabilities that are out there that folks are using to do that, that's what we're leveraging, so it's really more about sometimes I just need more capacity for the Pentaho server, but I don't need it all the time. Not every customer is going to get to the scale that they need that so it's a really easy way to just keep bringing in as much capacity as you need and have it available. >> David: I see, so really efficient, sort of low-level kind of stuff. >> Yes. >> So, when you talk about distributed load execution, you're pushing more and more of the processing to the edge and, of course, Brian gave a great talk about edge to outcome. You and I were on a panel with Mark Hall and Ella Hilal about the, so called, "power of three" and you did a really good blog post on that the power of the IOT, and big data, and the third is either predictive analytics or machine learning, can you give us a quick sense for our viewers about what you mean by the power of three and how it relates to pushing more workloads to the edge and where Hitachi Vantara is going in terms of your roadmap in that direction for customers. >> Well, its interesting because one of the things we, maybe we have a recording of it, but kind of shrink down that conversation because it was a great conversation but we covered a lot of ground. Essentially that power of three is. We started with big data, so as we could capture more data we could store it, that gave us the ability to train and tune models much easier than we could before because it was always a challenge of, how do I have that much data to get my model more accurate. Then, over time everybody's become a data scientist with the emergence of R and it's kind of becoming a little bit easier for people to take advantage of those kinds of tools, so we saw more of that, and then you think about IOT, IOT is now generating even more data, so, as you said, you're not going to be able to process all of that, bring all that in and store it, it's not really efficient. So that's kind of creating this, we might need the machine learning there, at the edge. We definitely need it in that data store to keep it training and tuning those models, and so what it does is, though, is if you think about IMS, is they've captured all that data, they can use the predictive algorithms to do some of the associations between customer information and the censor data about driving habits, bring that together and so it's sort of this perfect storm of the amount of data that's coming in from IOT, the availability of the machine learning, and the data is really what's driving all of that, and I think that Mark Hall, on our panel, who's a really well-known data-mining expert was like, yeah, it all started because we had enough data to be able to do it. >> So I want to ask you, again, a product and maybe philosophy question. We've talked on the Cube a lot about the cornucopia of tooling that's out there and people who try to roll their own and. The big internet companies and the big banks, they get the resources to do it but they need companies like you. When we talk to your customers, they love the fact that there's an integrated data pipeline and you've made their lives simple. I think in 8.0 I saw spark, you're probably replacing MapReduce and making life simpler so you've curated a lot of these tools, but at the same time, you don't own you're own cloud, you're own database, et cetera. So, what's the philosophy of how you future-proof your platform when you know that there are new projects in Apache and new tooling coming out there. What's the secret sauce behind that? >> Well the first one is the open-source core because that just gave us the ability to have APIs, to extend, to build plugins, all of that in a community that does quite a bit of that, in fact, Kafka started with a customer that built a step, initially, we've now brought that into a product and created it as part of the platform but those are the things that in early market, a customer can do at first. We can see what emerges around that and then go. We will offer it to our customers as a step but we can also say, okay, now we're ready to productize this. So that's the first thing, and then I think the second one is really around when you see something like Spark emerge and we were all so focused on MapReduce and how are we going to make it easier and let's create tools to do that and we did that but then it was like MapReduce is going to go away, well there's still a lot of MapReduce out there, we know that. So we can see then, that MapReduce is going to be here and, I think the numbers are around 50/50, you probably know better than I do where Spark is versus MapReduce. I might be off but. >> Jim: If we had George Gilbert, he'd know. >> (laughs) Maybe ask George, yeah it's about 50/50. So you can't just abandon that, 'cause there's MapReduce out there, so it was, what are we going to do? Well, what we did in the Hadoop Distro days is we created a adaptive, big data layer that said, let's abstract a layer so that when we have to support a new distribution of Hadoop, we don't have to go back to the drawing board. So, it was the same thing with the execution engines. Okay, let's build this adaptive execution layer so that we're prepared to deal with other types of engines. I can build the transformation once, execute it anywhere, so that kind of philosophy of stepping back if you have that open platform, you can do those kinds of things, You can create those layers to remove all of that complexity because if you try to one-off and take on each one of those technologies, whether it's Spark or Flink or whatever's coming, as a product, and a product management organization, and a company, that's really difficult. So the community helps a ton on that, too. >> Donna, when you talk to customers about. You gave a great talk on the roadmap today to give a glimpse of where you guys are headed, your basic philosophy, your architecture, what are they pushing you for? Where are they trying to take you or where are you trying to take them? (laughs) >> (laughs) Hopefully, a little bit of both, right? I think it's being able to take advantage of the kinds of technologies, like you mentioned, that are emerging when they need them, but they also want us to make sure that all of that is really enterprise-ready, you're making it solid. Because we know from history and big data, a lot of those technologies are early, somebody has to get their knees skinned and all that with the first one. So they're really counting on us to really make it solid and quality and take care of all of those intricacies of delivering it in a non-open-source way where you're making it a real commercial product, so I think that's one thing. Then the second piece that we're seeing a lot more of as part of Hitachi we've moved up into the enterprise we also need to think a lot more about monitoring, administration, security, all of the things that go at the base of a pipeline. So, that scenario where they want us to focus. The great thing is, as part of Hitachi Vantara now, those aren't areas that we always had a lot of expertise in but Hitachi does 'cause those are kind of infrastructure-type technologies, so I think the push to do that is really strong and now we'll actually be able to do more of it because we've got that access to the portfolio. >> I don't know if this is a fair question for you, but I'm going to ask it anyway, because you just talked about some of the things Hitachi brings and that you can leverage and it's obvious that a lot of the things that Pentaho brings to Hitachi, the family but one of the things that's not talked about a lot is go-to-market, Hitachi data systems, traditionally don't have a lot of expertise at going to market with developers as the first step, where in your world you start. Has Pentaho been able to bring that cultural aspect to the new entity. >> For us, even though we have the open-source world, that's less of the developer and more of an architect or a CIO or somebody who's looking at that. >> David: Early adopter or. >> More and more it's the Chief Data Officer and that type of a persona. I think that, now that we are a entity, a brand new entity, that's a software-oriented company, we're absolutely going to play a way bigger role in that, because we brought software to market for 13 years. I think we've had early wins, we've had places where we're able to help. In an account, for instance, if you're in the data center, if that's where Hitachi is, if you start to get that partnership and we can start to draw the lines from, okay, who are the people that are now looking at, what's the big data strategy, what's the IOT strategy, where's the CDO. That's where we've had a much better opportunity to get to bigger sales in the enterprise in those global accounts, so I think we'll see more of that. Also there's the whole transformation of Hitachi as well, so I think there'll be a need to have much more of that software experience and also, Hitachi's hired two new executives, one on the sales side from SAP, and one who's now my boss, Brad Surak from GE Digital, so I think there's a lot of good, strong leadership around the software side and, obviously, all of the expertise that the folks at Pentaho have. >> That's interesting, that Chief Data Officer role is emerging as a target for you, we were at an event on Tuesday in Boston, there were about 200 Chief Data Officers there and I think about 25% had a Robotic Process Automation Initiative going on, they didn't ask about IOT just this little piece of IOT and then, Jim, Data Scientists and that whole world is now your world, okay great. Donna Prlich, thanks very much for coming to the Cube. Always a pleasure to see you. >> Donna: Yeah, thank you. >> Okay, Dave Velonte for Jim Kobielus. Keep it right there everybody, this is the Cube. We're live from PentahoWorld 2017 hashtag P-World 17. Brought to you by Hitachi Vantara, we'll be right back. (upbeat techno)