 Live from San Francisco, California, it's theCUBE covering the IBM Chief Data Officer Summit. Brought to you by IBM. We're back at the IBM CDO conference at Fisherman's Warf and San Francisco. You're watching theCUBE, the leader in live tech coverage. My name is Dave Vellante. Glenn Finches here is the global leader of big data analytics at IBM. And we're pleased to have Mark Claire. He's the head of data enablement at AstraZeneca. Gentlemen, welcome to theCUBE. Thanks for coming on. Great to be here. Hi Mark, I got to start with this head of data enablement. That's a title that I've never heard before and I've heard many thousands of titles in theCUBE. What is that all about? Well, I think the credit goes to some of the executives at AstraZeneca when they recruited me. I've been a Chief Data Officer at several of the major financial institutions both in the U.S. and in Europe. AstraZeneca wanted to focus on how we actually enable our businesses, our science areas and our businesses. So it's not unlike a traditional CDO role, but we focus a lot more on what the enabling functions or processes would be. So it sounds like driving business value is really the main thrust? Well, it's three. I've always looked at this role in three functions, value, risk and cost. So I think that in any CDO role you have to look at all three. I think you'd slide it if you didn't. This one with the title obviously we're looking in quite a bit at the value we will drive across the firm and how to leverage our data in a different way. I love that because you can quantify all three. All right, Glenn, so you're the host of this event. So awesome. I love that little presentation that you gave. So for those who didn't see it, you gave us pay stubs and then you gave us a website and said take a picture of the pay stub, upload it and then you showed how you're working with your clients to actually digitize that and compress all kinds of things. Time to mortgage origination, time to decision. So explain that a little bit and what's the tech behind that and how are people using it? Sure. So for three decades we've had this OCR technology where you take a piece of paper, you tell the machine what's on the paper, what longitude and latitude coordinates there are and you feed it in and you hope and pray to God that it isn't in there wrong, the form didn't change, anything like that. That's the way we've lived for three decades. With cognitive and AI, I read things like the human eye reads things. And so you put the page in and the machine comes back and says hey, is this an invoice number? Hey, is this social security number? That's how you train it as compared to saying here's what it is. So we use this cognitive digitization capability to grab data that's locked in documents. And then you bring it back to the process so that you can digitally reimagine the process. Now, there's been a lot of use of robotics and things like that. I'm kind of taken existing processes and I'm making them incrementally better, right? This says, look, you now have the data of the process, you can reimagine it. However, in fact, the CEO of our client ADP said, look, I want you to make me a Netflix, not a Blutter blockbuster, right? So it's a mind shift, right? To say, we'll use this data, we'll read it with AI, we'll digitally reimagine the process and it usually cuts like 70 or 80% out of the cycle time, 50 to 75% out of the cost. I mean, it's pretty groundbreaking when you see it. So, Mark, as a head of data enablement, you hear something like that and you're not myopically focused on one little use case. You're taking a big picture view and you're doing strategies and trying to develop broader business cases for the organization. But when you see an example like that, and many examples out there, I'm sure the light bulbs go off. Oh, I wrote probably 10 use cases down while Glenn was presenting. We'll be talking about that later. You do get tactical, though, right? Okay, but where do you start when you're trying to solve these problems? Well, I look at the Glenn's example and about five and a half years ago, Glenn was one I went to, I'd gone to a global financial service firm and obviously having scaled across dozens of countries. And I had one simple request to Glenn's team as well as a number of other technology companies. I want cognitive intelligence on data ingest because the processes we've had done for 20 years just wouldn't scale. Not at speed across many different languages and cultures. And I now look five and a half years later and we have beginning of, I would say, technology opportunities. When I asked Glenn that question, he was probably the only one that didn't think I had horns coming out of my head that I was crazy. I mean, some of the leading technology firms thought I was crazy, asking for cognitive data management capabilities. And we are five and a half years later and we're seeing AI applied and not just on the front end of analytics but back in the back end of the data management processes themselves and starting to automate. So I look, there's a concept now coming out, data ops. And data ops is, if you think of what DevOps is, it's bringing within our data management processes, it's bringing cognitive capabilities to every process step and what level of automation can we do? Because for typical data science experiment, 80 to 90% of that work is data engineering. If I can automate that, then through a data ops process, then I can get the insight much faster but now I can scale it and scale a lot more opportunities and have to manually do it. So I look at presentations and I think in every aspect of our business where could we apply it? And what do you mean? What do you talk about data engineering? You talk about data science spending his or her time just cleaning data, wrangling data, all the not fun stuff. Plugging in cables back in the infrastructure days. I mean, you're seeing horror stories right now. I heard from a major academic institution, a client came to them and their data scientists that they had spent several years building were spending 99% of their time trying to cleanse and prep data. They were spending 90% cleansing and prepping and of the remaining 10%, 90% of that, fixing it where they fixed it wrong the first time. So they had 1% of their job doing their job. So this is I think a huge opportunity because you can start automating more of that and actually refocusing data science on data science. So you've been a chief data officer, number of financial institutions. You've got this kind of cool title now which touches on some of the things a CDO might do and you're a technical, you've got a technical background. So when you look at a lot of the, what Ginny Rometti calls it, incumbents. He called them incumbent disruptors two years ago at IBM Think. They've got data that has been hardened in all these projects and use cases and it's locked in people talk about the silos. Part of your role is to figure out, okay, how do we get that data out and leverage it, put it at the core? Is that fair? Well, and I'm going to stay away from the word core because to me core can infer kind of legacy processes of building a single repository, a single warehouse which is very time consuming. So I think of, can I leave it where it is but find a way to unify it? When I say core, I mean mentally the core of the business. That's what I need to think about is how to do this logically and create more of a unification approach that has speed and agility with it versus the old physical approaches which took time and resources. So that's a computer science problem that people have been trying to solve for years. Exactly. Decentralized distributed architectures, right? And why is it that we're now able to tap your technical expertise? I think it's the perfect storm of AI, of cloud, the cloud native, of IoT because some of you think of IoT. It's IoT to be successful as a fabric that can connect millions of devices or millions of sensors. So you pair those three with the investment Big Data brought in the last seven or eight years and Big Data to me initially when I started talking to companies in the Valley 10 years ago at the early days of MapReduce what I saw were companies and I can go to almost any of the digital companies in the Valley. They were using technology to be more agile. They were finding agile data science before we called it data science. The MapReduce and Hadoop were just almost, not an afterthought, but it was just a mechanism to facilitate agility and speed. And so if you look at how we've built that all the way up today and all these, all the convergence of all these new technologies, it's a perfect storm to actually innovate differently. Well what was profound about MapReduce and Hadoop it was like leave the data where it is and ship five megabytes of code to a petabyte of data and that, I think you're bringing up a good point. We've now, we've spent 10 years leveraging that at much lower cost and you've got the cloud now for scale and now machine intelligence comes in that you can apply on the data because as Bob Pitchiano once told me, data is plentiful, insights aren't, right? Amen to that, yeah. So okay, so this is a really interesting discussion. You guys have known each other for a couple of decades. How do you work together to solve problems? What is that conversation like? Do you want to start that? Yeah. So first of all, we've never worked together on solving small problems, not commodity problems. We would usually tackle something that someone would say would not be possible. So normally Mark is a change agent wherever he goes. And so he usually goes to a place that wants to fix something or change something in an abnormally short amount of time for an abnormally small amount of money, right? So what's strange is that we always find that space together and Mark is very judicious about using us as a services firm to help accelerate those things. But then also we build in a plan to transition us away and transition him into full ownership, right? But we usually work together to jumpstart one of these wicked hard, wicked cool things that nobody else really wants to tackle. So do people hate you at first or do they love you at the end of it? Yeah, I went into one institution and I said okay, we're going to do a four step plan. I'm going to bring the consultants in day one while we find talent internally and recruit talent external. So that's kind of phases one and two in parallel. And then we're going to train our talent as we find them. And then Glenn's team will knowledge transfer and by phase four we're running it. And that's a model I've done successfully in several organizations. People kind of hate it at first because they're not doing it themselves but they may not have the experience and the skills. And I think as soon as you show your staff that you're willing to invest in them and give them the time and the exposure, the conversation changes. But it's always a little awkward at first. I've run heavy attrition in some organizations at first to build the organizations but the one instance that Glenn was referring to, we came in there and they had a 12 to 15 year plan. And the CIO looked at me and says, I'll give you two years. I'm a bad negotiator. I got three years out of it and I got a business case approved by the CEO a week later that was a significant size business case in five minutes. I didn't have to go back a second or third time but we said we're going to do it in three years. Here's how we're going to scale an organization. We scaled more than a thousand person organization in three years of talent. But we did it in a plan way. And I mean, in that particular organization probably a year and a half in I had a global map of every data and analytic role I needed. And I could tell you where in the US they set and with what competitor or in what industry and where in India they set and in what industry and when we needed them we went out and recruited them. But it took time to build that but you know in any really period I've worked because I've done this for 20 plus years now the talent changes and location changes somewhat but it's always been a challenge to find them. I guess it's good to have a deadline I guess. You did not take the chief data officer role in your current position. Explain that. What's your point of view on that role and how it's evolved and how it's maybe being used in ways that don't line with it? I mean I think that a CDO and during the early days there wasn't a definition that matter of fact every time I get a recruiter call me oh we have a great CDO role for him. First thing I ask him is would you define what you mean by CDO? Because I've never seen it defined the same way in two companies. It's just that way. But I think that the CDO regardless of the institution as responsibility end in to make sure there's an end in framework from strategy execution including all of the governance and compliance components and that you have ownership of each piece in the organization. CDO in most companies doesn't own all of that but I think they have a responsibility in too many organizations that hasn't occurred so you always find gaps in each organization somewhere between risk costs and value in terms of how the organization's driving data. And in my current role like I said I wanted the focus we wanted the focus to really be on how we're enabling. And I may be enabling from a risk and compliance standpoint just as greatly as I'm enabling a growth perspective on the business or cost management and cost reductions. I mean I've been successful in several programs for self funding data programs for multi years by finding costs. I've gone into several organizations that had a decade of merger after merger and data is the afterthought in almost any merger. I mean there's a data silo session tomorrow and it'd be interesting to sit through that because I've found that data is the afterthought in a lot of mergers but yet I knew of one large healthcare company that made data core to all of their acquisitions and it was one of the first places they consolidated and they grew faster by acquisition than any of their competitors. So I think there's a way to do it correctly but in most companies you go in you'll find all kinds of legacy silos and duplication and those are opportunities to find to really reduce costs and self fund all the improvements, all the strategic programs you want to do. And I'm inferring from that in the data roll overlaps or maybe better than gaps in data is that thread between cost, risk and upside. It is and I've been lucky in my career I've reported to CIOs, I've reported to COOs and I've reported to CEO. So I've kind of reported in three different ways and each of those executives really looked at it a little bit differently. Value obviously is in a CEO's office. Compliance more in a COO's office and cost was more in the CIO domain but we had to build a program looking at all three. You know I think this topic though that we were just talking about is how these roles are evolving. I think it's natural because we're about five to seven years into the evolution of the CDO. It might be time for a CDO 2.0 and so you see more CDOs moving away from pure policy and compliance to more value enablement. It's a really hard change and that's why you're starting to see more turnover of some of the CDOs because people who are really good CDOs at policy and risk and things like that might not be the best enablers. So I think it's pretty natural evolution. Great discussion guys, we've got to leave it there. They say data is the new oil. Data's more valuable than oil because you can use data to reduce cost, to reduce risk, the same data to drive revenue and you can't put a gallon of oil in your car and a quarter of oil in your car, a quarter of oil in your house. So data we think is even more valuable. Gentlemen, thank you so much for coming on theCUBE. Thanks so much, a lot of fun, thanks. All right, keep it right there everybody. We'll be back with our next guest. You're watching theCUBE from IBM CDO 2019. We'll be right back.