 33 miles, 47 miles, 56 miles, 69 miles, 83 miles. It's never easy, but stopping isn't an option. Distance, 100 miles. From Fisherman's Wharf in San Francisco. It's theCUBE. Covering IBM Chief Data Officer Strategy Summit, Spring 2017. Brought to you by IBM. Hey, welcome back everybody. Jeff Frick here with theCUBE along with Peter Burris from Wikibon. We are in Fisherman's Wharf in San Francisco with IBM Chief Data Officer. Strategy Summit, Spring 2017. Coming to the end of a busy day, running out of steam, blah, blah, blah. I need more water. But Joe's going to take us home. We're joined by Joe Selly. He is the Global Operations Analytics Solution Lead for IBM. Joe, welcome. Thank you. Thank you very much. You've been in sessions all day. I'm just curious to get kind of your general impressions of the event and any surprises or kind of validations that are coming out of these sessions. General impressions is everybody is thrilled to be here. And the participants, the speakers, the audience members all know that they're at the cusp of a moment in business history of great change. And that is as we graduate from regular analytics which are more descriptive and dashboarding into the world of cognitive, which is taking the capabilities to a whole other level. Many levels actually advance from the basic things. And you're in a really interesting position because IBM has accepted the charter of basically consuming your own champagne, drinking your own champagne, whatever expression you want to use. I'm so glad you said that because most people say eating your own dialogue. If we were in Germany, we would talk about beer, but we'll stick with the champagne analogy. But really trying to build, not only to build and demonstrate the values that you're trying to sell to your customers within IBM, but then actually documenting it and delivering it basically, it's called the blueprint in October. We've already been told it's coming in October. What a great opportunity. Part of that is the fact that Ginny Rometti, our CEO, had her start in IBM in the consulting part of IBM, GBS, Global Business Services. She was all about consulting to clients and creating big change in other organizations. Then she went through a series of job roles and now she's CEO and she's driving two things. One is the internal transformation of IBM, which is where I am. Part of my role is, I should say, reporting to the chief data officer and the chief analytics officer. And their jobs are to accelerate the transformation of big blue into the cognitive era. And Ginny also talks about showcasing what we're doing internally for the rest of the world and the rest of the economy to see because parts of this other companies can do, they can emulate our roadmap, the blueprints rather, sorry, that Interpol introduced is going to be presented in the fall. That's our own blueprint for how we've been transforming ourselves. So some part of that blueprint is going to be valid and relevant for other companies. So you have a dual reporting relationship, you said. The chief data officer, which is this group, but also the chief analytics officer. What's the difference between the chief data officer, the chief data analytics officer and how does that combination drive your mission? Well, the difference really is the chief data officer is in charge of making some very long-term investments, including short-term investments, but let me talk about the long-term investment. Anything around an enterprise data lake would be considered a long-term investment. This is where you're creating an environment where users can go in, these would be internal to IBM or whatever client company we're talking about, where they can use some themes around self-service, get at this information, create analysis, everything's available to them. They can grab external data, they can grab internal data, they can absorb Twitter feeds, they can look at weather company information. In our case, we get that because we're partnered with the weather company. That's the long-term vision of the chief data officer is to create a data lake environment that serves to democratize all of this for users within a company, within IBM. The chief analytics officer has the responsibility to deliver projects that are sort of the leading projects that prove out the value of analytics. So on that side of my dual relationship, we're forming projects that can deliver a result literally in a 10 or 12-week time period or a half a year, not a year and a half, but short-term and we're sprinting to the finish, we're delivering something. It's quite minimally scaled. The first project is always a minimally project. It's using as few data sources as we can and still getting a notable result. So the chief analytics officer is at the vanguard of helping the business think about use cases, going after those use cases, asking problems the right way, finding data with effectiveness as well as efficiency and leading the charge. And then the chief data officer is helping to accrete that experience and institutionalize it in the technology that practices the people, et cetera. So the business builds the capability over time. Yeah, scalable. It's sort of an issue of it can scale. Once Interpol and the chief data officer come to the equation, we're going to scale this thing massively. So high volume, high speed, that's all coming from a data lake. And the early wins and the medium term wins maybe would be more in the realm of the chief analytics officer. So on your first summary a second ago, you're right in that the chief analytics office is going around and the team that I'm working with is doing this to each functional group of IBM, HR, legal, supply chain, finance, you name it, and we're engaging in cognitive discovery sessions with them. What is your roadmap? You're doing some dashboarding now, you're doing some first generation analytics or something, but what is your roadmap for getting cognitive? So we're helping to burst the boundaries of what their roadmap is, really build it out into something that was bigger than they had been conceiving of it. Adding the cognitive projects and then program managing this giant portfolio so that we're making some progress and milestones that we can report to various stakeholders like Ginny Rometti or Jim Cavanaugh who are driving this from a senior, senior executive standpoint. We need to be able to tell them in one case every couple of weeks what have you gotten done, terrible cadence by the way, it's too fast. But we have to get there every couple of weeks we've got to deliver another few nuggets. But in many respects analytics becomes a capability and data becomes the asset. Yes, that's true. Analytics has assets as well though because we have models and we have techniques and we bake the models into a business process to make it real so that people actually use it. It just doesn't sit over there as this really nifty science experiment. Right, right. But where are we on the journey? I think it's real still early, early days, right? Because we hear all the time about learning and deep learning and AI and VR and AI and all this stuff. Every organization is patchy, even IBM. What I'm learning from being here, so this is end of day one, I'm getting a little more perspective on the fact that we at IBM are actually, because we've been investing in this heavily for a number of years, I came through the ranks in supply chain, we've been investing in these capabilities for six or seven years, but we were some of the early adopters within IBM. But I would say that maybe 10% of the people at this conference are sort of in the category of I'm running fast and I'm doing things, so that's 10%. Then there's maybe another 30% that are jogging or fast walking and then there's the rest of them, so maybe 50% if my mouth is right. It's been a long day. Are kind of looking and saying I got to get that going at some point. And I have two or three initiatives, but I'm really looking forward to scaling it at some point. So I've just painted a picture to you of the fact that the industry in general is just starting this whole journey and the big potential is still in front of us. And then on the champagne, so you've got the cognitive, you've got the brute, and then you've got the Watson. And there's a lot from the outside looking in at IBM. There's a lot of messaging about Watson and a lot of messaging about cognitive. How do the two mesh and how do they mesh within some of the projects that you're working on? Or how should people think of it? People should know that Watson is a brand and there are many specific technologies under the Watson brand. So, and think of it more as capabilities instead of technologies. Things like being able to absorb unstructured information. So you've heard, if you've been to any conferences, whether they're analytics or data, any company, any industry, 80% of your data is unstructured and invisible. And you're probably working with 20% of your data on an active basis. So do you want to go to the 80% and shrink it? That's true. Yeah, because the volumes are grown. You're pulling in size, but shrinking as a percentage. So, just think about that. So is the Watson really the kind of the packaging of cognitive for a specific application? Yeah, Watson is a mechanism and a tool to achieve the outcome of cognitive business. That's a good way to think of it. And Watson capabilities that I was just about to get to are things like reading, if you will. Watson Health reads oncology articles. And once one of them has been read it's never forgotten. And by the way, you can read 200 a week and you can create the smartest doctor that there is on oncology. So a Watson capability is absorbing information, reading. It's in an automated fashion improving its abilities. So these are concepts around deep learning and machine learning. So the algorithms are either self-correcting or people are providing feedback to correct them. So there's two forms of learning in there. But these are kind of capabilities all around Watson. I mean, there's so many more optical character recognition right. Retrieven ranks are giving me a strategy and telling me there's an 85% chance, Joe, that your best move right now given all these factors is to do X. And then I can say well, X wouldn't work because of this other this other constraint which maybe the system didn't know about. Then the system will come, okay, well in that case you should consider Y and it's still an 81% chance of success versus the first one which was at 85. So retrieving and ranking these are capabilities that we call Watson. And we try to work those into all the job roles. So again, whether you're in HR legal, intellectual property management environmental compliance, regulations around the globe are changing all the time. Trade compliance and if you violate some of these rules and regs then you're prohibited from doing business in a certain geography, it's devastating. Mistakes are really high. So these are the kind of tools we want. So I'm just curious from your perspective you've got a corporate edict behind you at the highest level and your customers, your internal customers have that same edict to go execute quickly. So given that you're not in that kind of slow moving or walking or observing half what are the biggest challenges that you have to and even given the fact that you've got the highest level most senior edict behind you as well as your internal customers. Yeah, well guess what? It comes down to data often a lot of times it comes to data. We can put together an example of a solution that is a minimally viable solution which might have only three or four or five different pieces of data and that's pretty neat and we can deliver a good result but if we want to scale and Eddie sees and cares about our shareholder then we have to scale, then we need a lot of data. So then we come back to Interpol and the chief data officer role. So the constraint is one of the programs and projects is if you want to get beyond the initial proof of concept you need to access and be able to manipulate the big data and then you need to train these cognitive systems. This is the other area that's taking a lot of time and I think we're going to have some technology innovation here but you have to train a cognitive system. You don't program it you do some painstaking back and forth. You take your room full of your best experts in whatever the process is and they interact with the system. They provide input, yes, no they rank the efficacy of the recommendations coming out of the system and the system improves but it takes months. That's even a starting problem. That's a starting problem and then you trade it over an extended period of time. Well it gets better over time. You use this thing like a corpus of information is built and then you can mine the corpus. But a lot of people seem to believe and suddenly boom, you've got this new way of doing things and it is a very, very deep set of relationships between people who are being given recommendations, as you said waiting them, voting them, voting on them, etc. This is a highly interactive process. Yeah it is. If you're expecting lightning fast results you're really talking about a more deterministic kind of solution. If then, if this is then that's the answer. We're talking about systems that understand and they reason and they tap you on the shoulder with a recommendation and tell you there's an 85% chance this is what you should do and you can talk back to the system like my story a minute ago and you can say well it makes sense but or great, thanks very much Watson. I'm going to go ahead and do it. Those systems that are expert systems that have expertise just woven through them. You cannot just turn those on but as I was saying one of the things we talked about in some of the panels today was there's new techniques around training there's new techniques around working with these corp corpuses of information actually I'm not sure what the plural plural of corpus corpy. It's not corpy. Yeah somebody looked that up. More corpus? It's not corpy corpa anyway. So anyway I want to give you the last word, Joe. For you been doing this for a while what advice would you give to someone kind of in your role at another company who's trying to be the catalyst to get these things moving what kind of tips and tricks would you share having gone through it and been working on this for a while. Sure. I would, the first thing I would do is in your first moves keep the projects tightly defined and small with a minimum of input and keep, contain your risk and your risk of failure and make sure that if you do three projects at least one of them is going to be a hands down winner and then once you have a winner for your organization and then a lot of folks get so enamored with the technology that they start talking more about the technology than the business impact and what you should be touting and bragging about is not the fact that I was able to simultaneously read 5000 procurement contracts with this tool. You should be saying we used to take us three weeks in a conference room with a team of one dozen lawyers and now we can do that whole thing in one week with six lawyers. That's what you should talk about. Not the technology piece of it. Great. Well thank you very much for sharing and glad to hear the conversation going so well. Thank you and it's the answer to the question. All right, thanks Peter. Joe, Peter and Jeff, you're watching The Cube. We'll be right back from the IBM Chief Data Officer Strategy Summit. Thanks for watching.