 Live from San Francisco, California, theCUBE, covering MarkLogic World 2015. Brought to you by MarkLogic. Here are your hosts, John Furrier and Jeff Kelly. Okay, welcome back everyone. You are watching theCUBE live in Silicon Valley at MarkLogic World 2015. I'm John Furrier with my co-host Wikibon analyst, Jeff Kelly, our next guest is a customer of MarkLogic, Jeff Career Senior Director, product developed for Mitchell Juan. Welcome to theCUBE. Thank you. Jeff interviewed you on stage. So give us the lowdown. What's happening? What do you guys talk about? What's going on? What does your company do? And why are you here? Sure, Mitchell Juan's a subsidiary of Snap-on. Snap-on is a corporation very, very focused on just two things, right? Help people fix vehicles, help people run their shop better. At Mitchell, we do those things by delivering information to technicians on how to fix the car that's in front of them right then. Tell them what's wrong with it. From a shop management standpoint, we have a shop management software that's been running for 20 years and is now the source for our big data. So on day one, we threw the switch and had 20 years worth. So MarkLogic, CEO, is on earlier talking about the conversations that he's involved in. What are the top conversations that you're involved in with your job in terms of technology, your employees, customers, partners? What are some of the things you're talking about? How to take the information we have and better leverage it to fix cars? And where can we get new information? One of the things we're doing right now is we've got hundreds of millions of repair orders that have tons of information about what's going on in cars out on the road today. And what we're trying to do is figure out how to take that data and turn it into knowledge. To help a guy focus on the problem. He gets a car in his bay that he sees that car once a year maybe. He has no idea how to work on it. So what we do is try to help him by saying here's what we know about the car. We've seen it 170,000 times and here's the things that we know go wrong with it. And how to get that information to those people. We talk about that endlessly. So that's classic harnessing the data that already exists and put it in the hands of the user. Really fast. It sounds easy. I mean. Yeah, not that easy. Well, yeah. Well, you've got so many different sources of data we talked about in the panel. Talk about that a little bit. I mean, you're dealing with data coming from multiple sources in different formats. We have data coming from 27 different OEMs. And although the Society of Automotive Engineers would love to have you believe that there's a standard name for everything. Those 27 OEMs apparently didn't read that memo. So they will call, for example, an oxygen sensor by 27 different things. 27 different names. And a large part of what we have to do is to figure out what they mean in context. Again, to help that guy who doesn't know the 27 different names. But to him, it's an oxygen sensor. When they tell us that they then take that in the context of the vehicle they're actually working on and do it. So we have information coming in from OEMs. We have information coming in from hundreds of mechanics around the world. In North America, in Europe. If they see something unusual under the hood of a car that day that they're working on, they'll feed that information to us and then we incorporate it into our product and feed it to technicians. We're gathering sort of business intelligence, if you will, to the tune of millions of repair orders a month that we sort of pile on top of the 300 and some odd million we already have. And we're constantly chewing through that, looking for new things, looking for trends, looking for ways to help that guy fix the car. What I love about your story is that, I mean, you're truly dealing with big data and you're delivering it all the way to a frontline worker. Who's really, sometimes, John, we talk about business applications that are a little esoteric, a little kind of hard to define. But you're talking about a mechanic in a shop who's trying to fix a car and he needs certain data at a certain time. How is Mark Lodge helping you actually deliver on all that? What we've done is we've taken all the information from those 27 OEMs, we've put it into a Mark Lodge repository and basically broken it down into paragraphs. We enrich that with metadata about the context for that paragraph and so when somebody goes back to that oxygen sensor, we can say, well, here's a paragraph over here about how to remove it, here's a paragraph over here how to replace it. We assemble all of that into an article that's very, very focused on just that oxygen sensor and deliver that to their tablet, to their desktop PC, in theory to their phone, although that's a little suspect, right? Because it's a tiny little screen and a big wiring diagram doesn't look good on a screen that big. But very, very focused on getting them the right information at the right time to solve the problem that they have in their bay. It's one of the things I really love about the job is we're helping somebody do something, right? He's got a problem, we're helping him solve it. Yeah, they get faster turnaround on the inventory if they will, not sitting there, they get more churned, which is good for their business again, they make more money and they have happy customers. How hard is it? I mean, I was joking earlier, oh, it sounds easy. To unify all this data is really challenging. They remind the algorithms that need to be applied to the data. So it's a classic case of you have new inbound data coming in. And it might not be super fast, I mean, orders coming in, I mean, millions of month is significant. How real time? How do you unify the data? I mean, I know it must be difficult because I know the complexity of the data. Share with some of the nuances of how you guys unify it together and apply it. We have an editorial staff, total of about 50 people, some dedicated strictly to taxonomy, to ontology, to management of the vocabulary of all those OEMs, if you will. And so they're all touching that, they're all making sure that it's got the right markup, that it has the right metadata attached to it, that we're identifying new things because they're adding new things to cars constantly, right? So we can't kind of define a car and walk away. We've got to redefine that car every year, again, for the 27 different people. For the aftermarket, one of the big challenges we face, particularly with the repair orders, is technicians may be really good at fixing a car, spelling's not necessarily their strong suit, grammar's not necessarily their strong suit. So as we're doing the analytics on those repair orders, we've got to sort of be able to derive, oh wait, when they use this term, what they really meant was, or to look at things in context, and to cut out the noise, because nobody cares that that car got an oil change, right? It's not germane to a problem with the car, typically. And so just sorting the wheat from the chaff, and that's very challenging. That's more of the ontology side, sounds like. So you have developers then, working on this stuff. Yeah, so how does the interface with Mark positive, they get, what's their user experience like, dealing with the Mark logic technology? We've built a very nice front end, that lets them sort of manage the taxonomy, manage the ontology. We're doing some things now with what we call the knowledge graph, which is relationships across types of data. You know, when I talk about the 27 OEMs, they don't all use the same term for something. You know, their parts guys call it one thing, their labor guys will call it something else. The guy in engineering who wrote the service manual called it some third thing. And so stitching all of those together into a seamless experience for our users, it's very challenging and very demanding. We invest a lot in that. How about predictive analytics? Anything on prescriptive and predictive analytics you guys working on? Like, you must see things early and have to blast out notifications. We are, we're trending in that direction. I don't want to give away too much by way of private knowledge. The analogy I used earlier today, I don't know if you know the car talk on NPR, and a guy called up in the middle of winter from Maine and he gets one of the guys and he describes, he goes, yeah, I'm driving a 2008 Ford F-150 and one of the brothers goes thermostat. And the guy says, well, I didn't tell you the problem. And he goes, all right, tell me the problem. The guy describes this problem and he goes thermostat. And it turns out that, you know, what happens is a technician will be on one or two automotive forms. Well, the car talk guys have all day to do that, right? So they're looking across 40 or 50 of them. And what they had seen was this trend in sub-zero temperatures for Ford F-150 thermostats to go bad. And so as soon as the guy said, Ford wouldn't have 50 thermostat. So we want to do that same thing, right? We're from Maine, it's called record winter, Ford 150 connect the dots. Exactly. So we want to do that for our customers and it's a ways off here but we're heading that direction. And I think we've got a good start and we, you know, certainly competitive advantage just having 300 million repair workers because you're not running down to 7-11 and buying those when you decide you want to do a big data project. So walk us through kind of how you came to work with our project and one of the things that I'm interested in and you talked about it briefly in our panel discussion but maybe you could expand a little bit on some of the challenges of adopting a new technology and a new mindset. Not necessarily a technology challenge but a mindset challenge, you know, kind of giving, putting that relational model aside and adopting kind of a new mindset in this new world of big data. Talk a little bit about that and how that happened at your organization. Yeah, the mindset can be very challenging. You know, as I related earlier, when I came from Snap on Diagnostics which is where I had been and the Snap on Culture was very much we cut metal with fire, right, right. And so I get transferred to Mitchell and I think, all right, fantastic. I'm going to the information group and I get to the information group and they put ink on paper. It was like, no, fellas, really, there's no more paper. And getting people's head around stop thinking about relational technology just think about the problem you want to solve. Focus on the problem. Let the technology help you. You know, don't try to describe relationships between things. Mark logic will help us with that. You know, it lets us connect all of the dots so you don't have to manually do that anymore and we're still driving that change through. The technology side of the business, everybody gets it. The rest of the business is coming around but it's an interesting cultural transformation. And does that require investment in training, you know, existing workers so that they can adopt this or is it more of an informal kind of thing where employees help each other? I mean, how does it actually manifest itself? A bit of both. We encourage everybody very actively to take training. The Mark logic folks have done a great job of making the training accessible, making it affordable. So there's a lot of that but there's also a lot of helping each other. Part of it, no small part of it is shedding old bad habits because people think relationally and I've been around sadly long enough to remember pre-relational database, I remember relational database was the big new thing and everybody became very invested in it and getting people to sort of look past that can be an interesting challenge. We're getting there though and it's a lot of just talking people along, helping them understand, you know, stuff. Don't worry about the how, worry about the what, you know, and in my team I'll worry about the how in the background. So I got to ask you, what's the coolest thing that you're working on right now that you could share? I mean, you're involved in a lot of cool tech. Actually got a platform behind it in Mark logic and other technology. What is the, what gets you excited right now? What are the coolest, what's the coolest thing that you're doing and working on? Well, we just introduced a feature in one of our products that will show you the relationship between a symptom, all the parts of the car that might cause that symptom, all of the diagnostics codes, you know, the check engine light comes on so there's codes in between and we're showing a graphic relationship so that a technician can, you know, if the code comes up and he'll click on the code and we'll say, well, okay, here's a list of symptoms that we know are associated with that code because we did the analytics on the 300 million repair orders. And then if they pick one of those symptoms we'll go, aha, you know, it's that part right there is where you should start. And all of the combinations and permutations and it's very advanced and our competitors, I hope, hate me by first name, you know, because I want those guys going, I hate that guy because they did that cool feature. That is probably the coolest thing we're doing and I think we're just scratching the surface really with analytics. So talk about the enablement that you get from the Markologic and the Internet these days because the data you just liberate and you get to do more things, you get some creative licenses, great for developers. What kind of hiring challenge is out there? Obviously we hear all the time, it's really, really hard to hire full stack developers. So what are you looking for in the technology and what has helped you guys if you can share your experiences and getting good people, keeping good people? Because sometimes if people walk out the door that's homegrown, you're holding a bag. Two things is we look for learners, right? We don't necessarily insist that you come to us with Markologic skills. We insist you come to us with a willingness to learn. And then we try to keep the job interesting, keep the job fun, make sure that our developers are connected with their customer. I mean, one of the luxuries that we're afforded because there's aftermarket shops everywhere. And so all of our developers actually, I make sure that they get at least once a year and I try for twice to actually go out for a day and visit shops. We're a business that hires a lot of former technicians, a lot of guys that spent 10, 15, 20 years in shops. And so we'll send a developer out with them and they'll just go on shop visits and see how people are using the technology or using the application, you know. And we just ask them, what are you missing? What do you need? What are we not providing you that we could help you with? And that just keeps it interesting for everybody. Well, really appreciate you coming on theCUBE. Congratulations on being on the panel with Jeff Kelly, the people on Analyst, Big Data House, which his reports have been circulating around. He predicted this years ago, Jeff. Congratulations. Thank you so much. Great to have you on theCUBE. Thank you so much for coming on theCUBE. This is theCUBE, we'll be right back after this short break. To hear more from practitioners, industry leaders here on theCUBE in Silicon Valley from MarkLogic World 2015, we'll be right back.