 Live from Miami, Florida, it's theCUBE. Covering IBM's data in AI Forum, brought to you by IBM. Welcome back to Miami, everybody. You're watching theCUBE, the leader in live tech coverage. We're here covering the IBM data and AI Forum. Scott Buckles is here, to my right, he's the business unit executive at IBM and longtime CUBE alum Daniel Hernandez as the vice president of data and AI group. Good to see you guys, thanks for coming on. Hi, so we're going to talk about, you're very welcome. We're going to talk about data ops, kind of accelerating the journey to AI around data ops, but what is data ops and how does it fit into AI? Dan, you'll start with you. There's no AI without data. You've got data science to help you build AI. You've got DevOps to help you build apps. You've got nothing to basically help you prepare data for AI. Data ops is the equivalent of DevOps, but for delivering AI-ready data. So how are you, Scott, dealing with this topic with customers? I mean, is it resonating? Are they leaning into it? Are they saying what? No, it's absolutely resonating. We have a lot of customers that are doing a lot of good things on the data science side, but trying to get the right data at the right people and do it fast is a huge problem. They're finding they're spending too much time prepping data, getting the data into the models, and they're not spending enough time failing fast with some of those models or getting the models that they need to put in production into production fast enough. So this absolutely resonates with them because I think it's been confusing for a long time. So AI's scary to a lot of people, right? It's a complicated situation, right? And how do you make it less scary? Talk about problems that could be solved with it? Basically, you want a better customer experience in your contact center. You want a similarly amazing experience when they're interacting with you on the web. How do you do that? AI is simply a way to get it done and to get it done exceptionally well. So that's how I like to talk about it. I don't start with, here's AI, tell me what problems you could solve. Here are the problems you've got. And where appropriate, here's where AI can help. So what are some of your favorite problems that you guys are solving with customers? Customer and employee care, which basically is any business that does business has customers. Customer and employee care, huge problem space. Catching bad people. Financial crimes investigation is a huge one. Fraud, KYC-AML as an example. National security, things like that, right? You spend all your time with customers, what else? Well, customer experience is probably the one that we're seeing the most. The other is being more efficient, helping business solve those problems quicker, faster, try to find new avenues for revenue, how to cut costs out of their organization, out of their runtime. Those are the ones that we see the most. So when you say customer experience, immediately chatbots jumps into my head. But I know we're talking more than, it's sort of a transcendent chatbots, but double click on customer experience. How are people applying machine intelligence to improve customer experience? Well, when I think about, if you call into Delta and you have one bad experience or your airline, whatever that airline might be, that customer experience, or that could lead to losing that customer forever, and there's used to it don't be an old adage that you have one bad experience, you tell 10 people about it, you have a good one, you tell one person or two people. So getting the right data to have that experience is where it becomes a challenge. And we've seen instances where customers, or excuse me, organizations are literally trying to find the data on the screen while the customer is on hold. So they're saying, can I put you on hold? And they're trying to go out and find it. So being able to automate finding that data, getting it in the right hands of the right people at the right time in a moment's notice is a great opportunity for AI and machine learning. And that's an example of how we do it. So from a technical standpoint, Daniel, you guys have this IBM CloudPak for data and this kind of magic data virtualization thing. Take an example that Scott just gave, but think of an airline. When I'm on my, I love my mobile app. I can do everything on my mobile app, except there's certain things I can't do. I have to go to the website. There's certain things I have to do with e-commerce that I have to go to the website that I can't do. Sometimes watching a movie, I can't order a movie from the app. I have to go to the website, the URL, and then order it there and put it on my watch list. So I presume there's some technical debt in each of those platforms. And there's no way to get the data from here and the data from here talking to each other. Is that the kind of problem that you're solving? Yes, and in this particular case, you're actually touching on what we mean by customer and employee care everywhere. The interaction you have on your phone should be the same as the interaction and the kind of response on the web, which should be the same, if not better, when you're talking to a human being. How do you have exceptional customer and employee care, all channels? Today, say the art is, I've got a specific experience for my phone, a specific experience for my website, a specific different experience in my contact center. The whole work that we're doing around Watson Assistant and it as a virtual assistant, is to be that nervous system that underpins all channels. And with Compact for Data, we could deliver it anywhere. You want to run your contact center on IBM Cloud, great. You want to run it on Amazon, Azure, Google, your own private center or everything in between, great. Compact for Data is how you get Watson Assistant, the rest of Watson and our data stack, anywhere you want, so you could deliver that same consistent, amazing experience, all channels, anywhere. And I know the tone of my question was somewhat negative, but I'm actually optimistic, and there's a couple examples I'll give you. I remember Bill Belichick one time said, ah, the weather, it can't ever get the weather right. This is probably about five, six years ago. Actually, they do pretty well with the weather, compared to 10 or 15 years ago. The other is fraud detection. I mean, in the last 10 years, fraud detection has become so much better in terms of just the time it takes to identify a fraud, and then the number of false positives, even the last, so I'd say 12 to 18 months, false positives way down. I think that's machine intelligence, right? I mean, if you're using business rules, they're not way down, they're still way up. If you're using more sophisticated techniques that are depending upon the operational data to be trained, then they should be way down. But there's still a lot of these systems that are based on old school business rules that can't keep up. They're producing alerts that in many cases are ignored, and because they're ignored, you're susceptible to bad issues. With, especially AI-based techniques for fraud detection, you better have good data to train this stuff, which gets back to the whole data ops thing, and training those with good data, which data ops can help you get done. And a key part of data ops is the people in the process. It's not just about automating things and automating the data to get it in the right place. You have to modernize those business processes and have the right skills to be able to do that as well. Otherwise, you're not going to make the progress. You're not going to reap the benefits. Well, that was actually my next question, is what about the people in the process? We were talking before off camera about RPA and me saying pave the cow path, but sometimes you actually have to re-engineer the process and you might not have the skill set. So it's people in process and then technology you'll lay in. And we've always talked about this, technology's always going to change. Smart technologists will figure it out, but the people in the process, that's the hardest part. What are you seeing in the field? We see a lot of customers struggling with the people in process side, for a very variety of reasons. The technology sends me the focus, but when we talk to customers, we spend a lot of time saying what needs to change in your business process when this happens? How do those business rules need to change so you don't get those false positives? Because it doesn't matter at the end of the day. So can we go back to the business rules thing? So it sounds like the business rules are sort of an outdated, policy-based, rigid sort of structure that's enforced no matter what, versus machine intelligence, which can interpret situations on the fly. But can you add some color to that and explain the difference between what you call the business rules-based versus AI-based? So the AI-based ones, in this particular case, probably classic statistical machine learning techniques, they do something like know who I am. My name is Danny Hernandez. If you're to Google Danny Hernandez, the number one search result is going to be a wrapper. This is a wrapper that actually just recently came out. He's not even that good, but he's a new one. A statistical machine learning technique would be able to say, all right, given Daniel and the context information I know about him, when I look for Daniel Hernandez and I supplement the identity with that contextual information, it means it's one of the six that works at IBM, right? And usually- Not the wrapper. Not the wrapper. Not the wrapper. Exactly. I don't mind being matched with a wrapper, but match me with a good wrapper. All you have to do is search Daniel Hernandez in the cube and you'll find him. That's right. Bingo, actually that's true. So, in any case, the AI-based techniques basically allow you to isolate who I am based on more features that you know about me so that you get me right. Because if you can't even start there, with whom are you transacting, you're not going to have any hope of detecting fraud. Or either that, you're going to get false positives because you're going to associate me with someone that I'm not. And then it's just going to make me upset because when you should be transacting with me, you're not because you're saying I'm someone I'm- So that ties back to sort of what we were talking about before, know your customer and anti-money laundering. Which of course was big, it still is, during the crypto craze. Maybe crypto's not as crazy, but that was a big deal when you had Bitcoin at whatever it was. What are some practical applications for KYC, AML that you're seeing in the field today? Cool. I think that what we see a lot of is, where we're applying in my business, is automating the discovery of data. And learning about the lineage of that data, where did it come from? This was a problem that was really hard to solve 18 months ago. Because it took a lot of manpower to do it. And as soon as you did it once, it was outdated. So we've recently released some capabilities within Watson Knowledge Catalog that really help automate that so that as the data continues to grow and continues to change as it always does, that rather than having 2,300 business analysts or data stewards trying to go figure that out, machine learning can go do that for you. So all the big banks are glomming onto this? It's global. We think about any customer onboarding, right? You better know who your customer is and you better have provisions around anti-money laundering. Otherwise there's going to be some very serious downside risk. It's just one example of many, for sure. Let's talk about some of the data challenges because we've talked a lot about digital, digital business, and I've always said difference between a business and a digital business is how they use data. So what are some of the challenging issues that customers are facing? Particularly incumbents. Ginny Rometti used the term a couple of events ago. Might have been even World of Watson. Incumbent disruptors. Maybe that was the first thing. Which I thought was a very poignant term. So what are some of the data challenges that these incumbents are facing and how is IBM helping solve them? For us, one of them that we see is just understanding that where their data is. There's a lot of dark data out there that they haven't discovered yet. And what impact is that having on their analytics? What opportunities aren't they taking advantage of? And what risks are they being exposed to by that being out there? Unstructured data is another big part of it as well. Structured data is sort of the easy answer to solving the data problem. But unstructured data. But still hard. But still hard. But unstructured data is something that almost feels like an afterthought a lot of times. But the opportunities and risks there are equally, if not greater to your business. Yeah, so you're saying it's an afterthought because a lot of times people say that's too hard. Right. Just forget it. But there's gold in them in our hills, right? Yeah, absolutely. So how does IBM help solve that from using tooling? Is this discovery tooling or are you? Well yeah, so we recently released a product called InstaScan that helps you to go discover unstructured data within any cloud environment. So that was released a couple months ago. That's a huge opportunity that we see where customers can actually go and discover that dark data, discover those risks, and then combine that with some of the capabilities that we do with structured data too. So you have a holistic view of where your data is and then start tying that together. If I could add, any company that has any operating history is going to have a pretty complex data environment. Any company that wants to employ AI has a fundamental choice. Either I bring my AI to the data or I bring my data to the AI. Our competition demand that you bring your data to the AI, which is expensive, hard, often impossible. So if you have any desire to employ this stuff, you had better take the I'm going to bring my AI to the data approach or be prepared to deal with a multi-year deployment for this stuff. So I think that's just that principle difference in how we think about the problem means that we could approach problems, we could help our customers apply AI to problem sets that they otherwise couldn't because they would have to move. And in many cases, they're just abandoning projects all together because of that. So now we're starting to get into sort of data strategy. So let's talk about data strategy. So it starts with, I guess, understanding the value of your data. Start with understanding what you got. Yeah, what data do I have? What's the value of that data? How do I get to that data? You just mentioned you can't have a strategy that says, okay, move all the data into some God box. Good luck. Yeah, that won't work. So do customers have coherent data strategies? Are they formulating? Where are we on that maturity curve? Absolutely, I think the admin of the CDO role as the chief data officer role has really helped bring the awareness that you have to have that enterprise data strategy out there. So that's a sign. There's a CDO in the house. You know somebody's working on enterprise. Yeah, absolutely. It's really their role. The CDO's role is to construct the data strategy and support it. And one of the challenges that we see, though, and that is that because it's a new role is going back to Daniel's historical operational stuff, there's a lot of things that you have to sort out within your data strategy of who owns the data, regardless of where it sits within an enterprise. And how are you applying that strategy to those data assets across the business? And that's not an easy challenge. That goes back to the people process side of it, for sure. Well, right. I mean, I bet you if I asked Jim Kavanaugh, what's IBM's data strategy? I bet you he'd have a really coherent answer. But I bet you if I asked the Scott Hebner, the CMO of the data and AI group, I bet you I'd get a somewhat different answer. And so there's multiple data strategies, but I guess it's Inderpal's job to make sure that they're coherent and tie in, right? Absolutely. Absolutely, absolutely. Quick study. So what's IBM's data strategy? Data is good. Data is good. Bring AI to the data. Look, I mean, the data topic, I mean, so data and AI, that's the name of the business, that's the name of the portfolio that represents our philosophy. No AI without data. Increasingly, not a lot of value of data without AI. We have to help our customers understand this. That's a skill education point of view problem. And we have to deliver technology that actually works in the wild, in their environments, not as we want them to be, but as they are. Which is often messy. But that's, I think that's our fun. It's the reason why we've been here for a while. All right, I'll give you guys last word we've got to run, but both Scott and Daniel, takeaways from the event today, things that you're excited about, things that you learn, give us the bumper sticker. For me, you talk about whether people recognize the need for a data strategy and the role. For me, it's people being pumped about that, being excited about it, recognizing it, and wanting to solve those problems and leverage the capabilities that are out there. So you're seeing a lot of that today. Absolutely, and we're at a great time and place where the capabilities and the technologies with machine learning and AI are applicable and real that they're solving those problems. So I think that gets everybody excited, which is cool. It's fun to see. Excitement, a ton of experimentation with AI, some real issues that are getting in the way of full-scale deployments, a methodology, data ops, to deal with those real hardcore data problems in the enterprise, resonating, a technology stack that allows you to implement that as a company is through Cloudpack for Data, no matter where they want to run, is what they need and I'm happy that we're able to deliver it to them. Great, great segment, guys. Thanks for coming on. Thank you. Data, applying AI to that data, scaling with the Cloud, that's the innovation cocktail that we talk about all the time on theCUBE. Scaling data your way, this is Dave Vellante and we're in Miami at the AI and Data Forum brought to you by IBM. We'll be right back right after this short break.