 Live from Las Vegas, expecting the signal from the noise. It's theCUBE, covering InterConnect 2016. Brought to you by IBM. Now your host, John Furrier and Dave Vellante. Okay, welcome back everyone. We are here live in Las Vegas for exclusive coverage of IBM InterConnect 2016. This is theCUBE, it's our flagship program. We go out to the events and extract the signal from the noise. I'm John Furrier, my co-host Dave Vellante. Our next guest is Chris O'Connor, General Manager of Watson, Internet of Things offering at IBM. Welcome to theCUBE. Thank you. Obviously IOT is hot, we were just talking about it on our intro. And it really is a great enabler because now you have the data, you have edge devices now anywhere. But there's a huge, it's not just sexy in the sense of it's a hot trend, there's real business value there. And this is a huge buzz. I mean, it's the froth on the market. People are talking about the next big thing because of the value is in the billions and the TAM basis, it's huge. Hundreds. Hundreds of billions of dollars from savings, new revenue. So the dream can be sold. That's right. So, Jack, what's the starting point? I mean, you guys have new, you've got some great success. Talk about some of the results in your group. And the state of what's going on. So the starting point is fairly ubiquitous. Everybody's starting from this point of operational value. And if I can get the data, I can bring it in. I can take whatever I'm doing today, the process I'm using to run that thing today, I can do it better now that I know about it in real time. So I can instantly take today's people and today's processes and I can make them more efficient. I can make them operate better. And you can often pay for the program of instrumenting just off the operational savings that goes along with it. So that buys the program right there. And once you're instrumented and connected, you can go after a couple of other things. You can go after warranty and lifecycle. You can push that down your supply chain. Start to talk to your suppliers different because you're instrumenting what's coming into. More important, you can talk to your distribution chain. And you can look at what they're actually doing with your product when it goes through its lifecycle out to whoever eventually uses it. And you can push a different dynamic in how you work with them. And then most important, you can actually talk to your customers. So take one of our clients like Whirlpool. They're able because of the instrumentation they're going to do with the Internet of Things for the first time to actually talk directly to the end user of their refrigerator or the dishwasher or whatever they put in your home because they're going to disintermediate that middle layer of the department store whoever sold you the machine. And they're going to be able to see you directly. So not only are they going to get better warranty and operational cost, they're going to talk to the client. And so that's really cool for companies that have been physical oriented to now be able to digitally talk to the client. That's right. So there's low hanging fruit that sounds like and this easy wind you could knock down. That's right. Pace for itself, get a starting point. There's no, you know, throw the Hail Mary, build a day will come kind of mentality because you are have stuff connected to the network. So the first wave is whatever's in the asset network or in your ownership. That's right. And then you go from there. What is the value then? Take us through some of those examples because this is where people kind of go, okay, I just want to start. So give us some examples. So if you have to fix something, if you make something that's physical and you have to fix it. And that thing is existing in different locations around the world. That means you have to do one thing, which you have to roll a truck. So you've got to send somebody out there with a part and how often do they have the right part? Do they have the right part for the right model to be able to fix that? And it is at the right person that goes. And so you got to roll the truck one point something times if you're good or you got to roll it more than that if you're not the savings right there. When you think about somebody that makes appliances that makes 32 million appliances under warranty today, car manufacturer might have 25 million cars at any point in time rolling around the road under warranty. And you think about what they can start to control by understanding diagnostically by the information coming in, what to do. They can operate that much better. Take a home appliance. When you call in with a broken home appliance, there is logically no question they can ask you that will actually help you fix your appliance. They only have one choice, which is, is it plugged in? Is it cold? Is it hot? Or did it clean the dishes? And then after that, they roll a truck. So now with diagnostics, I can tell you stuff like that the water to your house is at low pressure. You don't have any more hot water. We show a drain clogged indicator. So let me show you how to do that. And you can talk somebody through it and you can avoid that whole discussion. And that's assuming that those endpoint devices are those assets if you will, have connectivity. That's right. So where do we, where do you see this sort of connectivity spectrum? Are the windmills, do they all have internet? So the thing that's changed, when we first started doing Smart Cities 10 years ago as IBM, we literally had to think about putting the sensors in the devices and then we had to think about connectivity. And in those cases, like when we did the city of Stockholm, we ran wires, right? We dragged physical wires to the video cameras through the streets, under the streets, right? Now it was barbaric, right? When you think about how we connected this stuff, right? Today, if you go look at it, Wi-Fi is ubiquitous, it's widespread. It's campus, it's city-wide as an option. You've got all these terraced versions of cell phone connectivity that the telcos have to keep alive as well. So you can go with all different levels of telco connectivity. And then you have specialty options like Laura and others that are out there. So connectivity is cheap, it's very available. If you look at the chip suppliers and the folks that- And it's getting more pervasive as well. That's right. It's built into all the boards now. It's just an assumed component. When people go throw together a board that there's going to be a connectivity, whether it's Bluetooth to a local gateway, or whether it's something more broadband oriented that they can get the data and they get the information. So what are some of the conversations, before we get into the Watson tech side, which is a cool insight aspect of it, what are some of the conversations? Because I'll see the value that you're talking about. You can knock down some stuff quick. You have some sort of vision. You can see some value down the road. Are people mindful of this? Is it kind of like, are they scratching their heads? Oh my God, it's going to more work for me. Or are they excited? What's the orientation of the buyer that you talk to? Because, I mean, assuming they're excited, but are they excited? Are they a little bit challenging? What are some of the conversations? The transformation that's taken the past two years is it's gone from theoretical exploration oriented to clients in every industry now looking at the connectivity of a thing as a way to fundamentally operationalize today's cost and then drive some sort of wedge of transformation that they think their competitors won't have. Insurance industries looking at how they can buy into better policy and management of different types of casualty, property life with instrumented things being able to augment their actuarial table oriented view of how they do things to healthcare providers being able to send patients home with instrumentation that they wear, that gets them out of the hospital sooner and actually provides them a more connected experience than the four hour check that you get from your nurse to the industrial examples that we talk about around the car. So it's disruptive to them, but they see it as a positive disruption. That's right. For the most part, right? They see it as a positive disruption and the clients we're working with see they have no choice and in fact if they want to win in their market they've got to get there first. They get that, yeah. That's right. So at the high level, how are they measuring the business impact? It seems like there's a return on the assets because they're instrumenting the assets and it seems like there's a productivity piece, revenue per employer, how have you measured that? How are people measuring it at a high level? So the metrics do vary, but just to give you a couple of the times, so the insurance industry, for example, has no excuse to talk to the clients they've already signed up. So the internet of things and the ability to drive this instrumentation gives them an excuse to go in and re-talk to their clients and re-monetize the value that they're already providing them. So they're actually measuring the ability to go and drive client-acceptant rate and the increased number of clients that they have around the policies. So for them, that's a measurement they can see and they can track. If you go over to some of the folks that are doing life cycles around healthcare, they're actually looking at patient days in hospital, they're actually looking at patient recovery rates and they actually look at satisfaction of delivery services. So re-admittance. Re-admittance, all those types of things become measurable statistics that come out there. So each industry, we can see the metrics that actually beyond just operationalizing pennies better, which when you have a lot of pennies that adds up, but they can actually drive business value at the same time. And they also see the opportunity to disrupt the supply chain. If I can tell you what you bought and what you left behind in the dressing room because of the internet of things, knowing you and the inventory you brought into there, I can go back to my suppliers and I can give them direct feedback versus a fall-spring bicycle. And I can give them direct feedback on the type of product that they need to make and what's selling and the fact that everybody took that one thing in the dressing room, they all said, I'm not buying that. And then it's immediate feedback that they can drive into the supply and distribution chain as well. That's where Watson starts to get interesting. So you create a learning environment out of the whole thing. So you're dealing with scale of thousands, tens of thousands, millions in almost all of these cases. And this is exciting. The Watson thing is the insight engine. So it's the cognitive, that's the buzzword. We're talking about pre-cognitive is having the data set up so you connect, so you instrument the devices, get the data, right? Once you get the data, you bring it to Watson and then Watson does its thing. So we get that. I want to ask you a specific question around Watson. In your successes you've had this year, is there something you can point to or share anecdotally that's a process improvement that's changed the outcome? Meaning, we hear stories of, oh, big data was an outlier that we couldn't have got to before and we've saved a billion dollars on them. It's one trained company guy interviewed. Literally, they were doing stuff on spreadsheets and they saw this weird outlier that they would have regressed out with the modeling and changed a billion dollars in savings. So that's huge. Is that kind of stuff going on? That type of stuff's going on. There's safety examples that come to light. We're working with folks that deal around shipping containers and they're able to take the internet of things and from a safety and a security point of view, they're using the patterns of how things move around in there to be able to analyze for safe and good practices versus fraud and they're able to find not only millions of dollars of savings in goods that disappear and don't happen to show up, but they're also able to, by analyzing these patterns and then learning from the patterns, they're able to even go after elicit activity at the same time. There's savings that take place around not just savings, but there are examples around being able to take how people move as you instrument elevators and people movers. And being able to understand not only is the device working correctly, but they're actually now looking at the people inside and learning what the behavior is of them and then being able to make changes in terms of how that device works with you in terms of moving you to the right destination. Elevators as a service could be a solution out there. Someone might come up and do stuff like that. Okay, so process improvement. So they're actively taking the data and injecting and changing process. Example there. So let's take the example we announced with Kone. Largest elevator people mover in the world, they claim to move a billion people a day. They'd like to tell me they move the planet every seven days, right? So great Finnish institution and they are instrumenting their elevators. And what they see out of that is 400,000 locations, heavy machinery, custom installed in many cases. And so when they go to roll and go do that, and there's also a regulatory environment that goes along with it at the same time. So you ever get an elevator and you look at the little sign off sheet down there and they have it signed off and you wonder who gets in this thing and signs it off, right, every week. Expired? Right, and different countries have different regulations around that as well. They're able to go back and actually challenge the regulatory environment in different countries now as a result of the internet of things and be able to show that they can certify the digital sign off versus the mechanical inspection. And for them, that's a tremendous savings. It's actually a safer environment by using the digital sign off they find than having the human interface out there. And they're able to drive a change in the industry as a result. There's real data on that. There's real data. It's not just supply chain per se. It's all kinds of value chain. That's right. So the cognitive IoT ecosystem feels like it's an infinite playground. Can you sort of describe what's going on there, the partnerships that you're going after and how do you prioritize them? There's just so many opportunities. So first there's a phenomena that's kind of a reversal of classic analytics. We spent 20 years teaching everybody in the data center that you should reduce your data. Get your data, reduce it, get it down, right? Run analytics, get it down. Cognitive analytics, cognitive IoT is about increasing the amount of data that you actually process to have a better learning environment. So if you have a thing that's running down the road, you're bringing in the information on the thing, but you also want to combine that with other environmental vectors that are out at the same time, such as the weather. So if you... It's not a database issue. Just database. You have to merge. You have to merge. Just pair a data source. And learn. So our IBM fellow used a brilliant example in his speech. I don't know if you saw it the other day, but he talked about when a baby in a high chair drops their bottle, that baby learns about gravity. When a car slips on the road, all cars should learn about the slip. That's cognitive intelligence at work. It's not just about the single thing, but it's about connectivity to the environment and then about propagating the learning environment out to instantly be able to benefit everything that's associated with that same type of instance. There's a ways telling me about a pothole. There's maybe a better way. The cars... You usually hit it by that point, right? Thank you. I hit the pothole. I was too busy putting my ways input into the car while driving. That's right. So, In-Heech, you saw was on yesterday. She's now running the collaboration group over there and she was kind of talking about the research center around one of that. Get the place. She was talking about this, got all the displays and she mentioned that the people who built it did the minority report. So we were kind of riffing on that last night and this morning we... And we were kicking around this term pre-cog, which is the people who sat in that pool and predicted the minority report, the crime thing if you've seen the movie, you know what I'm talking about. But we're trying to find a way to just talk about the pre-cognitive insights. Meaning, before Watson takes the ball, if you will, or takes the data, is there a pre-text to all this? Obviously collecting the data. Can you help us? I mean, what do you call it? Is it pre-cog or is it what pre-prep? I mean, data prep has always been out there. But like, what do you guys talk about? Standard techniques through analytics that help take streams of information and make sure that it's clean and valuable, apply the right rules of it, get pushed out to the edge, help make sure that the edge is clean, that what comes in from those devices is ready to be consumed. You want to make sure that you stage through this discussion. You can't just bring all the data back. You have to bring the data back in orderly amounts. And when you combine it, you want to combine it for value. So there is staging of information that is things we've done for years around analyzing, reducing that stream, being able to make sure it's clean, being able to have rules around it, being able to process it, being able to mart it. But then you want to combine it with other data streams. And that's where the cognitive value comes in, which is merging multiple streams together, often from different sources of variants that give you that learning environment. But that's somewhat antithetical to the last five years of how we've done Hadoop, which is no scheme on right. Throw it into the data, lake, swamp, ocean, whatever it is. You're saying there's got to be some notion of schema before you're taking data in from the edge. Is that right? When you look at the proliferation of IoT devices and you listen to the numbers that people throw around, there's going to be 30 billion, there's going to be 25, I don't know what the right number is, but it's huge, right? You will want to do some edge processing. And you'll want to think about being able to get that information at the right level and clean before you chuck it into the lake and then combine it with other information that you want to process at the same time. We were at a Facebook developer conference in Silicon Valley. Went down to the Cube there. And a couple of different kind of observations. And I'm going to take the kind of tactic of being a skeptic, so I'm not necessarily a skeptic, but I'll put my skeptic hat on for a second. Whoa, Facebook pointed out, and the geeky conversations go in and saying, there's so much data streaming, there's a lot of streaming technologies, machine learning and all that. It's hard, you're going to miss something. So we know it's a math problem. There's math involved in all this in the analytics space. So I got velocity, like Facebook, I'm going to miss something, or I got trickle effect with whether it's UDP or HTTP protocol coming in off devices and I'm trying to do some data cleaning, I might miss something. That could be critical data. I'm nervous, I'm not going to move. That's FUD that caused me to stop. How do you guys answer that? Because I know Watson has done some things. How do you guys manage that objection in the motions with your customers? So, and I love the way you said it, because it kind of gives me the most easy way to tee back up the answer, which is the value and the patterns that hit necessitate the need to move now. You can't study the end-by-end problem forever because you can bring together a combination of the streams today and you can see immediate value. The example I gave you about the car and the slippage and the weather. How do devices work in combination with the physical conditions that they're on? What do humidity, temperature variations do? Metal shrinks and contracts. It runs different. Rubber reacts in different ways to the environment that's in. What is your velocity of business effectiveness? What's the business effectiveness you're going to have given these variants you bring together? And do I need to plan for more or less fuel? For example, based on how I'm going to see the traction of my fleet as it rolls down the road. Those are things that become real business calculations that say why there are streams you may still want to go after. You can't miss the fundamental value that's on the table today. So you're saying the betterment of the whole of the outcome is better because you get the value from the current learning. And I'd assume that the second answer might be just thinking if I was in your shoes would be and tell me if this would be an approach is the learning system is learning. That's right. So is that how you guys look at it? It is. We think that you need to start down the learning path. You've got enough variants to bring together today that you can then start that learning process and we can bring in other vectors. That's what machine learning conversation geeky under the hood stuff happens. That's right. And you want to find the anomaly that says you've seen this four times before. You don't realize it but you've seen this happen before. Sooner you stream this so you can get patterns. Sooner you get patterns as soon as the patterns can get better that's kind of the iteration. Is that? That's right. And underneath this is operational efficiency. I can't stress every client we start with is getting huge gains from operational efficiency today. They are paying for the program of instrumentation just off of operational efficiency. And so then you kind of It's so low hanging fruit. Just pick it. Just go get it. Just talk about your auto learning piece there. So a customer's got limited budget. I think they can focus on data, acquiring data, more data outside data sources, et cetera. And the other piece is improving the machine learning. That's right. So where are you seeing people put their resource? I mean obviously both. But how are they deciding to turn those knobs? In terms of the value investment they make first? Do I invest in gathering more data, finding more data sources, cleaning more data? The data effort because it's significant. It's a heavy lift. Or do I focus on just improving my algorithms and improving the machine learning? It seems like sort of too related but there are different efforts, different skill sets within customer bases. The first value tier we see right now given where the industry is is people are making sure they have instrumented capabilities and that they're bringing the information into a place where they can actually make organized sense out of it. That lake or that capability where I can start to see it relationed out against time. And they're spending energy in making sure that they have that definition correct. And then they're going for algorithmic value on top of it as the next place that they go build data. So it's somewhat linear right now. I want to ask about security. Yeah. Because the perimeter is blown away. The surface area is increasing. Yeah, right. There's no more data center. There's no virtual data center wall, right? So talk about how you guys are approaching that problem, how customers are dealing with that. So security falls into a set of patterns. And we help customers break it down into patterns. A one-way device is a completely different discussion than a two-way device. So a device that just sends something up and that tells you who it is, needs a certain set of information. And then you have to take in the pattern of what type of device is it doing, right? So do I want to set it up in a stream where there's actually some sort of hash that makes the device very authentic? Or am I really collecting a lot of independent little sensor data where I'm actually going for an aggregate view anyway and an independent node being corrupted is not a problem? So you help clients talk about the pattern of where they are and the type of devices and what it is that they're ingesting. And then when you go to a two-way communication stream, this is why you see IBM announcing things like blockchain. Like blockchain for IoT we think is a huge advancement and being able to have a device disconnect, go away from the world, come back and then rejoin its transactional stream. Where it's a known entity that is rejoined and because of that blockchain algorithm that's running in the background, its integrity is transactionally secure in terms of how it rejoins. So these types of discussions take place based on the pattern of who you are. In that example, there's no third-party verification. I mean, I hate to say it, but I mean it's conceptually anyway not hackable. It's not hackable. And there's significant value coming from each step along the way that you could choose to add extra layers. The chip vendors all have a vaulted oriented discussion in terms of how they think about being able to ID and fingerprint who they are. So you could build that into the discussion as well. It becomes part of your algorithm in terms of the blockchain. And you can work up the stack. So it's pattern-based. The other thing that I wanna point out comes into this is you get worried about privacy as you do this at the same time because when that car drives from Germany into France, you need to actually change where the data is recorded from being recorded in a German data center to being recorded in a French data center as well. And this is where we stand on top of the IBM cloud. We stand on the work that we're doing with software. 44 different data centers around the world. And that's important because you can actually manipulate the geo. That's right. Kind of like how IP addresses can be manipulated. So we actually work with clients today and some of the work we're doing with people like Kone is to be able to geographically segment the data based on the laws, regulations and rules of those clients. And that's a cool, I mean, Chris, we could talk with actually the whole day on this. It's a great topic and very relevant and very cool in my opinion, I think I love it. But we gotta move on. Maybe we can spend some more time at IBM Vision on this. That's a big show there. Again, that's coming up as IBM Vision is your show. But talk about your business, your opportunity. What are you guys doing with the go to market? What events you guys got coming up? What are you guys doing in the field? How are you guys engaging customers? What's your plan the next couple of months or so? So we're working on two primary initiatives. One is we're out meeting and working with all the major industries that are out there around helping them build out their solutions. Looking at customers that see solution stacks like the work we've done with Siemens and Kona and others. But we're really trying to enable a developer, I should say and, we're trying to enable a developer transformation at the same time. Go out to ibm.com slash IOT. You can register for a free Raspberry Pi, right? So we've gone out with Premier Fresnel and we've reoriented the Raspberry Pi board. Like hackathons stuff, get down and dirty. That's right. And so we're in hackathons. You can go and get your own Raspberry Pi. You can play with it at home. It costs about $20 on top of Amazon. We're giving away a thousand of those here at the show. So you register, you can get your own Raspberry Pi take. I have one of those golf things that Epson was showing. Actually, Dave and I got one. M-Tracer. Can't wait to use the M-Tracer on my golf swing that I think I have. I think I have a good swing, but I'm afraid to look at the data. They'll watch out for the tour you're coming, right? My internet of thing value proposition. Internet of job. Operationalize my swing. Chris, thanks so much. Internet of things is super important. Again, the low hanging fruit's so low. Why wouldn't you do it? Take it down right now. Fund it. Get the data. Learn. Plug in Watson. I'm assuming Watson's all big part of the message. That's right. Absolutely. Thanks for coming on theCUBE. Really appreciate the insights. Check out Twitter and go search CubeGems for all the highlights and the CUBE interviews. You can go to siliconangle.tv for all the videos. This is theCUBE. We are here live, exclusive at IBM Interconnect, bringing you all the action, all the digital assets, all the IoT, internet of things, the people, the content, the stories. We're here live at theCUBE right back.