(energetic music) >> Hey, welcome back, get ready. Jeff Frick here with theCUBE. We're at our Palo Alto Studios for a CUBEConversation, get a little bit of a break from the conference madness which is in full force right now. And we're excited to have our next guest, he's Mike Tuchen, the CEO of Talend coming off a really good quarter. Mike, great to see you. >> Thank you, Jeff. >> You guys are on fire! >> You know, it's a great time to be in the data business right now. (Jeff laughs) >> So give us a little update, what's going on recently? You've got a big show coming up, I imagine there's lots of announcements that are going to come out that you probably can't tell us about at the show, but go ahead and give a plug. It's coming up really soon and let's get into it. >> Yeah, exactly. So just in a couple weeks Talend Connect in London on the 15th and 16th and Talend Connect in Paris on the 17th and 18th. And Talend Connect is our user conference, so we'll have hundreds of people there, a lot of partners there, we'll roll out a whole bunch of new product announcements and talk about a lot of the great stuff that our customers are doing with Talend. >> So you've got an interesting way to kind of package up what you guys do in a really simple way and that's, you said before we turned on the cameras, the first mile, you know there's always so much conversation about the last miles, not necessarily in data but you know, in getting cable to your home and broadband and this, that, and the other but you talked about the first mile. Arguably, that's a lot more important than the last mile. >> Well you can't even get started on anything else until you solve the first mile problem and that's what we do. And the problem is, right now, every single customer in the world is waking up to the power of data and they need to be data-driven. They know it can make a huge difference in their business, and competitively the market leaders are all incredibly data-driven and if companies aren't equally data-driven then they fall behind. And so there's an incredible surge of interest in data-driven, becoming data-driven right now. The challenge that everyone faces is in order to get started down that path, your data is locked up in a lot of different places, it's dirty, it's inconsistent. And until you bring it together, clean it up, and make it consistent you can't do anything with it. That's the first mile. That's what we do. >> So how does it change now? I mean there's obviously been EDL and data cleansing issues for a very, very long time. So when you look at some of the trends, the growth of public cloud, obviously the explosion of data. Now you guys are taking a little bit different approach than kind of the historical method so how do you do it differently and why is it so important? >> From our perspective, we made a bet about five years ago when I joined, that the entire landscape, the IT landscape was being reinvented from the ground up. Not just the data world, the data world for sure, but the entire IT landscape was being reinvented. And that meant you had to solve the problem differently. And so from our perspective, there's four or five big trends that are completely reshaping the IT landscape. Number one of course is the move to the cloud. You've talked about it just a second ago but we're probably 10 years into a 20 or 30 year shift to the cloud and it's actually accelerating right now. We're now seeing not just early adopters but mainstream companies are now making a big bet on the cloud and deciding that's where they're going to be for the foreseeable future. We're seeing the move to more and more self service, where rather than having an IT team solve all your data problems, they're seeing data analysts and data scientists are solving the problems themselves. So creating a world where all of those different roles can play together in a team sport kind of way is an important way. It's moving to more and more real time, right? Everything back 10, 20 years ago used to be done in batch. So at the end of the day, or end of the week, or end of the month you collect a whole bunch of stuff and package it together and crank it through. But think about today's applications, right? The expectation is it's done in real time. If you make a deposit in a bank, you expect to look up the bank balance and see it right there. You don't expect to see it there the next day. >> Right, right. >> You expect your apps to be immediately responsive, that's real time, right? It's now this ubiquitous expectation. And that means that data integration needs to follow that. Tightly connected with that is the move to machine learning. Companies now don't want to do all of the analytics and insight generation with a whole bunch of people looking at data. 'Cause machines can do that a whole lot better, right? Machines are really, really good at finding patterns. And so those are some of the big trends that we see that are completely reshaping the landscape. So clearly, data integration today is just very different than where it was five or 10 years ago. >> It's so funny, we go to a lot of shows and there's always a lot of conversations about innovation and how do you innovate? And to me, one of the really simple answers, not necessarily simple to implement, is you give more people in the organization more access to more data and the tools to manipulate it and then ultimately, hopefully, to make decisions you know, based on that output. So it is kind of unlocking it, it is giving more people that access you talked about. Self service and cloud and really pushing that out and the other funny thing, you talked about real time, is you used to make decisions based on a sample of things that happened in the past. Now with the capacity of the machines, the complete, basically infinite capacity from an individual company point of view of a cloud application, now hopefully, I'm making decisions on all the data while it's happening. Completely different way. >> Yes, yes. And as a matter of fact, the outliers sometimes are really an important part of the data. And so looking at not just where does most the data fall, but why are the outliers there? What do they mean, right? In a fraud detection case the outliers are the frauds, usually, right? So it's an important part of the data and looking at the entire data set allows you to find that. If you're looking at a sample, you'll miss it. >> So as we look forward to machine learning, kind of the last part of your four key drivers, that's a big impact on the way these things work. My favorite little example on machine learning in AI is the new Google Gmail, on that little tiny response that it suggests that on your reply. Which seems relatively straightforward, right? Thanks, I'll get right back to you, you know they're relatively short usually. But the amount of machine learning and artificial intelligence and data analysis that goes into the generation of my three responses versus your three response options back to me is pretty phenomenal. And you guys are now going to be able to bake that into all types of general type processes. >> That's right, and that's right. You know, you described a really cool consumer scenario around e-mail, but there's a bunch of commercial scenarios around things like predictive maintenance. GE, with its big gas turbines. If that thing goes offline at the wrong time it can be real expensive. Because then you have customers who are out of service and it turns out it takes hours to spin up a new gas turbine that might be sitting idle. But if you can do it in a maintenance window it's just not a big deal at all. And so if they can predict when parts are about to fail, that's a savings of literally billions of dollars across their install base. We have one of the major car companies did a really cool analysis around predicting potential recalls based on, in manufacturing, as tools were starting to go out of alignment. And what they could do was start to track and say, if it gets more than this far out of alignment the odds of a recall go up dramatically, and so now's the time to intervene and readjust that tool because a recall is a very, very expensive thing. If you can fix it upfront in the tool you're saving millions of dollars. >> Right. >> Fascinating examples of real world industrial scenarios using machinery. >> Right, and disconnected kind of data sets that actually are tied together in hindsight but probably the person who's responsible for keeping that machine up and running isn't really thinking about the impact to the company if there's a recall on that particular model of car. >> Yeah, exactly. Who would have known that the tolerance, you know, acceptable tolerance was exactly this, right? How would you set that in advance? But it turns out when you actually start running the correlations and throw some learning algorithms at it, you can really start pinpointing it and say, for this tool, it's this, for this other tool it might be something else. >> So the other kind of big trend that you did mention in this explosion of data is using so many more data sets. Going beyond the data that you own, that you generate, that you create, and pulling in a lot of this external data whether it's weather data, whether it's social sentiment data. There's so many data repositories now that you can integrate in with that proprietary data to then drive kind of a secret sauce algorithm that gives you that competitive advantage. You see more and more of that and I think you mentioned kind of the sloppy, crazy variability in all these data sets as you're trying to pull them into these systems. >> That's right, that's right. And we're seeing a bunch of customers doing that. There was an interesting scenario of, we have a customer that does soil testing for farmers with a neat, little device, kind of an IoT scenario, they plug it in, it does the soil test, sends it up to the cloud. Now it correlates that soil with the weather patterns in that area to say, here is the seeding and fertilizing regimen that we should be using for this plot of land. Right? Really cool scenario. >> Well, I'll tell you even a crazier version. I talked to a guy that ran a drone company with the sensors that did a similar type of thing. They run the drone and they analyze the field. And I had to ask him, I'm like, "Come on, I mean people have been sampling fields forever, "this can't be new, right?" And then it feeds back to their little Monsanto engine or whatever that tells you what to do. He goes, "Yeah, but here's what's different, Jeff. "Again, we used to take a sample. "We would take sample points on that field "and we would make a decision based on that sample." He goes, "Now I can track literally every single plant." >> That's cool. >> "Every single plant with the consistency of this "drone coverage and now I can micro, micro, micro "the application of water, the application of hydrogen," or whatever they give, the herbicides, et cetera. Pretty amazing. >> Yeah, and what we're seeing now is that the tractor companies are doing that on a, as you say, on a per seed basis as they're driving through the field based on samples that have been taken, based on drone surveys of what's there, and based on the weather patterns. I mean it's really cool what we're doing in terms of precision farming right now. >> Right. So I'll just take that kind of one step further. The other trend that's coming down the pike which is big and not going to have less data but a lot more is IoT. So from where you're sitting you've been in this business a while, as you look at kind of this next generation of explosion of all this additional machine-generated data, what type of future do you see? How is that going to play? What kind of opportunities is that going to open up? There's a whole nother, multiple orders of magnitude of data coming soon. >> Yeah, no, so IoT is clearly a... It multiplies the amount of data by literally an order of magnitude of, and many of the streams are real time in nature and the absolute requirement then is that you're doing some sort of machine learning to take advantage of it. To me, you can take almost any industry and talk about a potential machine learning scenario in the industry. My favorite one right now is cars, right? This was, you know, it's now, it's in real life. It's not a future thing. If you're driving a Tesla right now, your car is actually starting to fix itself sometimes. Literally, I got a call one time as I was driving down the road, they say, "Hey, we've detected this fault in your car "and if it's okay with you we're going to reset it "right now and it'll be fine." And I was like, "What was the problem?" They're like, "Don't worry about it." Well, that's pretty cool, right? When was the last time-- >> Did they at least ask you to pull over first? (Jeff and Mike laugh) >> But no, the whole idea of having a car that's self-diagnosing and fixing itself is really cool. That's a game changer, I think. >> On so many ways, I mean not only that but you generalize that to a much broader audience. I mean it used to be you made your product, you sent it to your distributor, and you maybe had some assumptions of how it's used, how it's not used. Are people using the features that you created? Are they not using them? Are they using them they way you thought? And now with this connected feedback loop, the ability for manufacturers to know how people are using their tools even beyond just the prescriptive maintenance is a phenomenal impact. >> Yes, and in that particular scenario for those kind of smart devices, not just the one-way feedback loop, but closing the loop and the in field update ability is you know, you combine those two, and wow! It's a whole new world. >> Right, I guess software really is eating the world. I guess you had it right way back when. All right, Mike, well thanks for stopping by. Good luck on your event across the pond here in a couple weeks and great to catch up. >> All right, thank you, Jeff. >> All right, he's Mike, I'm Jeff. You're watching theCUBE. It's a CUBEConversation in our Palo Alto Studios. Thanks for watching and we'll see you next time. (energetic music)