AI Transformation: Why Your Data Landscape is Probably "Shocking"

Speaker A: Welcome to the Tech People Podcast. My name is Ken Coyne. I'm your host and founder, as well as an ambassador for OPS talent. I believe at the heart of any success story are the people who made it happen. Diversity, creativity and innovation, where nurturing people can lead to an unbeatable formula. I created this podcast to share the experiences of some truly inspirational leaders on a journey to success. Enjoy the show.

Speaker B: We've all heard the hype of AI, that it's going to revolutionize everything from customer service to backend operations. But here is a reality check most people aren't talking about. If your data landscape is a mess, AI isn't going to fix your business. It's just going to amplify your existing problems. I'm your host, Ken Coyne, and welcome back to Tech People, the show where we strip away the buzzwords to look at how technology actually works in the real world. Joining me today is Mark Rotwell Books, a partner at Verity X who is based in Dubai. Mark has spent years on the front lines of digital transformation and he has a bit of a controversial take. He believes digital transformation as we know it, it is dead. And the new era of AI transformation, which requires a brutal honesty about our data that, uh, many organizations just aren't ready for. Today we're diving into why most data landscapes are, in Mark's words, shocking. How to move AI, uh, out of the innovation Lab and into the real world, and how you can start seeing actual business value in months, not years. So with that, let's welcome Mark to the show.

Speaker C: Hey, Mark, great to have you on the show today.

Speaker D: Hi, Ken, how are you?

Speaker C: I'm doing very well. And how was life in sunny Dubai?

Speaker D: Well, as it's certainly sunny, that's for sure. Um, great time of year this time of year. It's not too hot and, yeah, it's paradise. Another day in paradise, as someone said. Yes, all good. How's things with you?

Speaker C: Not as hot. I'm in Belgium at the moment. A bit wet, a bit dull. A bit. Listen, could be worse at the same time, but tell us a bit for the audience. Tell us a bit about you, your background, and maybe how you ended up in Dubai and then we can go from there.

Speaker D: Yeah. So I'm technology background, originally. Experience working in large enterprises, banks primarily, but had consulting firms for about 20 years with the same theme, all about managing change and executing change swiftly and efficiently. So what we're doing at Veritx in the Gulf is on a similar theme. We have effectively four elements to what we do Dependent upon where you are as a customer. If you're curious. We have uh, a labs environment to enable you to innovate and fail fast and go again a little bit further on in your understanding of what you want to do. We have a core execution capability. So you know, you've decided you're going in on a change program, a change path that will help you execute that. And then we have a scale element of what we do, so we help you industrialize those changes. And then we have a regulatory uh, capability where we're sort of influencing and trying to get a regulatory alignment and helping define the new relationship between the regulator and the regulated. So we've been in the, been in the Gulf for about a year. Very positive so far. Lots of change going on, digital and AI change going on in this region. So that's why we're here, to satisfy that demand.

Speaker A: Yeah.

Speaker C: And how, I mean in terms of tech they are, they're very advanced at a kind of behind the curve or the middle ground. How would you describe it?

Speaker D: It's a bit of a mixed bag I'd say. I mean we deal quite a lot, quite extensively in banking. But you know, of, of late, since we've been here, we've also, we get involved in some government activity. But if you take, you know, digitally served in banking in the Gulf, I would say per capita is less advanced from a digitally served perspective than perhaps other regions. So there is still this um, you know, desire in banking, you know, to go and visit a branch which is something that in Europe and North America it's less so. But the, on uh, the reverse of that, you know, the strides in AI and the desire to move forward at pace in that regard is probably at the cutting edge of what people are doing on a global basis. So I'd say it's a mixed bag. It depends which angle you're coming from really. I think there is still a desire to complete on that digital transformation journey. And in a way it's sort of been leapfrogged by the desire, the bubble if you like, that is the AI transformation. So yeah, there's lots to do, that's for sure. And the ambition in the region is wonderful to see. You know, they do really want to push forward on um, in the vanguard of where, you know, where AI is going to help organizations and society more generally. So that's good to see. And then there's a healthy conflict is probably the wrong word, the healthy competition between the nation states in terms of who's actually who's going to take the lead there. So that's interesting to observe as an outsider in the middle of it sounds

Speaker C: like lots, uh, of great opportunities.

Speaker D: Yeah, there's certainly that. There's certainly that. As I said, you know, AI seems to be the thing that's everyone's banging on about at the moment.

Speaker C: Yeah, well, let's talk a bit further about that.

Speaker B: Right.

Speaker C: So I know you've mentioned in the past, and I quote you on this, digital transformation is dead, but now it's all about AI transformation. Talk to us about this. What's the distinction?

Speaker D: Well, I mean, that's a little bit of a provocative comment, so probably deliberately so. What I mean really by that is that digital transformation has been around now for 10 years. I think, you know, if you don't have a digital transformation program or you've not been pursuing that, you're probably out of business. So, you know, it is being done, has been done. Um, they are. I think the debate is still open as to whether or not it achieved the things it set out to achieve, but we can probably cover that in a bit more, greater detail a bit later on. But notwithstanding that, I think now that the industry seems to be now pursuing an AI transformation agenda. Whereas, you know, 10 years ago, the whole industry was driving towards the digitization of core processes and core capabilities, and the experience from a consumer perspective was being pushed down a digital channel for reasons which were valid at that time. Now we're seeing a lot of investment, effort, investigation, proof of concepts around how AI can be introduced at the corporate level for the benefit of the corporates and their customers. So when I say it's dead, I don't literally mean that. I mean, there are still digital transformation endeavors ongoing. And, um, certainly in this region, for the reasons I said, a little bit behind compared to other regions, one could say. But with regards to the AI journey and the AI train, you know, there's lots of people boarding that particular train and it's leaving the station very quickly.

Speaker C: Yeah. You know, what I find interesting is that, you know, gone through the whole digital transformation where we're still ongoing. Some companies, other companies have gone through it. Moving to the AI space. What amazes me is, you know, it's all about having the foundations in place. Right. Because you can't. Everyone wants to jump to the agents, but unless you've got a really strong fundamentals, strong foundations in place, you can't jump to that. And this is the, I suppose the big aspect we're talking about today is the whole area of data. So before, obviously the podcasts. We had a chat and you, you mentioned, you know, companies asking, okay, where do we start? How do we implement AI? Uh, and you said, well, first of all, let's look at your data landscape. How is that? And you said, quite often you get back shocking, which, uh, amazes me, you know, and it's still in the world. Here we are in 2026 and Dave, you think will be quite advanced, but still we're back. The basics. Uh, yeah, what's up with that?

Speaker D: That's very true. I mean, and I guess it's the rush, the rush, the rush. Let's understand how AI, how AI can help. The question I always respond when, you know, I get presented with that challenge was, okay, so what does your data landscape look like? Because in a way, it's all about the data, right? So, you know, and to the point about the digital transformation actually deliver what it set out to do, there's a very strong argument to say it probably didn't, you know, it just digitized inconsistent and incoherent processes. So, you know, it didn't really move the dial that everyone thought it would. My fear is that that is exactly what AI will do. I mean, it's an amplifier. So if you've got a bad data landscape or a bad process, all it's going to do is make that worse. So in my head, the solution is, if you're looking for where to start, assuming, of course, that organizations know the business case and the business rationale for implementing some of these AI solutions, that you need to start with those basics and have a look at the data landscape. Because a lot of organizations, when they have that introspective moment and they look back into their organizations, they're dealing with legacy systems and, you know, departments that have come in and been bolted on, but not really integrated, and you've got a, uh, data landscape that is very fragmented, not consistent or coherent against a strategy which is barely existent. If you look at a global strategy for your data landscape, for example, and you know, I would advocate, that's a really good place to start before you start understanding how you're actually going to utilize that data and get the value from that data with AI. So in my head, AI is all, you know, is the utopian position, and I get that, and I'm sort of sold on that Kool Aid as well. But I think unless you sort out those basics and those fundamentals, you're on a hiding to nothing. All you're going to do is replicate the traps that we fell into in the digital transformation days where we're just transforming bad processes and bad data landscapes and just making it worse. So I don't wish to be a downer on this. Unless you can work out how you're going to rationalize that data or where the value is in that data, then, you know, you do that first alongside a proper understanding of what the business case is for AI. Before you have technologists saying, okay, well, this is a great AI tool, let's implement this.

Speaker C: So that's a great point. Yeah, I think it's a great point to mention here. It's about where is the value in your data? You know, because. Yeah, well, I suppose what I'm seeing. I'm, um, sorry. You're seeing the same as you're seeing these big massive data projects that go on for months and if not years, building out whatever it is, data warehouses, data lakes. It's a huge investment. And what's the return on that? You know, it's challenging. Are you seeing the same M in the Gulf, these big data projects? For sure.

Speaker D: I mean, there's a lot. I mean, taking banking and, you know, the government areas, right. So they've had a legacy of no different. Every regions, I think, has gone through this, had a legacy of growth, if you like, you know, not that in the government space, growth by acquisition is a thing per se, but it's the same concept as a bank in terms of, you know, when you go and acquire a department over here or entity over there, the government departments have grown in the same way. And as a result of that, you get pockets of department X and department Y both having a HR system, both having, you know, some form of fulfillment system, both having a CRM system, slightly different systems, or could be the same technical system, but configured in a slightly different way. And the data doesn't speak. So there's lots of examples of that. It's certainly in banking, you know, from a legacy point of view and in government as well. So that is quite rightly pushing a desire to rationalize all of that, which, you know, obviously is the right thing to do to provide that, uh, stable ground from which you can build. But those projects and those initiatives take a while to unpick depending upon how problematic and where your starting point is. And we advocate. There has to be a middle ground there. I'm not saying you shouldn't rationalize your data landscape and have a proper strategy and go through a process of simplifying that and, um, with that, possibly a simplification of the application landscape. But I think you need to also accept that, you know, if that's a 12 or an 18 month endeavor for a large government department or a large bank, which is not unusual in terms of a timescale perspective, that's a big investment in capital, big investment in time. And meanwhile the business wants some value from this effort.

Speaker C: Exactly.

Speaker D: And it's not agile. So you know, you gotta find ways of introducing that agility, introducing that value whilst you're going through a process of rationalization and providing yourselves with that baseline and that foundation.

Speaker C: And are you finding ways to fast track these things? I mean, trying to get value quicker?

Speaker D: Yeah, I mean we, you and I have spoken about various tools where you can very quickly, relatively quickly provide yourself with the analytics and insights within the data whilst you are ongoing with a big data rationalization process. You know, without having to invest in team of data scientists, which, you know, let's be honest, once you've identified who they are, will probably take you three months to bring them into the organization. Another three months for them to understand what the hell's going on within your own data world and then possibly another three months for them to start making some inroads. So you know, that's one way of approaching it, which is nine months into your 18 month journey for data rationalization. And that doesn't really provide you with a solution. So there are tools which you could implement which, which enables that value to be brought a lot sooner by rationalizing some of your existing data landscape in a fragmented form, but putting some sanity towards it so you can then understand and derive some value whilst these big data programs and data scientists endeavors are ongoing. And that's the middle ground I talk about. I know that's uh, you know, an organization that would embrace that approach as a precursor, if you like to, you know, perhaps implementing some AI tools on the back of the value you can derive quickly within that data space whilst also running in parallel. You know, you're more embracing strategic data rationalization process.

Speaker C: Yeah, so what I'm saying, what I'm getting from this basically is okay, you kind of have to go through this pain one way or the other. You have to get your foundations in place with the data, build your data warehouse, get your data into one location and then you can start looking at the AI piece. But at the same time there is tools like Expanse AI, which is tool we use, right?

Speaker B: Yeah.

Speaker C: Which automates the role of the data scientists and get your value very quickly, but you still need to put those big foundations in place.

Speaker D: Yeah, I think there's space for both those things. Right, the foundational stuff, yeah, it's obvious, put it in. But accept that's a longer, medium term pain that you're just going to have to go through because there's no magic wand you can wave. And you know, unless you're a complete startup and you don't have this legacy. But there aren't that many organizations that are in that position, you know, and the Expanse AI example is a good example because that enables you to take what on the face of it is a fragmented landscape of data, uh, and consolidate that relatively quickly into a place where you can start doing that data analytics capability to derive that value whilst all the other things are ongoing. And that then provides with those data insights that you can then apply AI models to that to deliver that value a lot quicker than having to invest and wait 12 to 18 months before you can actually get something sensible out of your program of change. And when we talk about accelerating change, it's that sort of thing and it's that sort of approach that we would advocate for sure.

Speaker C: And talk to me about, um, let's go into those AI models, right. You mentioned earlier about you have the labs and I'm guessing you do a lot of PLC's proof of concepts. But how, I mean, are we still at that stage or are companies now starting to move past that? Or uh, what's the.

Speaker D: No, we're still. There's a lot of proof of concepts going on. Right. So there aren't that many referenceable, I would say operational implementations of AI certainly in the Gulf. Don't know about other regions we focus on the Gulf, but there aren't that many, you know, implementations of AI that you can put your hand on and say, actually okay, these guys have done this for this reason and this is the value that they're deriving. Now there are examples of that. So there is a bank that we did some work with where they were having an issue with the end of month reconciliation on FX and uh, they were finding that difficult and the regulator got involved as to why that was so problematic. And you know, there was a negotiation and a solution put forward that the regulator was comfortable with and that helped that reconciliation process operate and execute a lot more efficiently with AI as opposed to humans, obviously humans still in the loop from a checking point of view. But ultimately the heavy lifting was done by an AI model. So that is an example of true value. You know, you shorten the end of month reconciliation process. It's a lot more accurate, less prone to human error and something that the regulator understood, endorsed and was happy with. That's a good news story. But the few and far between, we're still in lots of proof of concept and phases and, you know, iteration and failing is a good thing, right? Because ultimately you're going to get to a point where you've iterated and succeeded and I think you've got to go through that process and not be afraid to try and fail and go again. And there's nothing wrong with that process inherently, but I think we are still in that stage, but I think we are coming out of that. I think there are point solutions where people are looking to productionize, if you like, AI proof of concepts. But some of the larger banks that we deal with, they've, uh, clearly set directions to say we want to implement AI in the context of service operations, for example, in an IT context. Not quite sure 100% what that means, but we've got a massive estate, loads of false positives, we've got oodles of people running around trying to interpret this stuff. We feel it's becoming a bit unwieldy and the solution is not a load more humans doing that sort of stuff. There's got to be a better way. But not quite sure. We've also got a reference architecture about how we want to implement that, but let's try it. So they're still in that phase and that's just one example, AI in the context of IT service operations. So I think we've got a bit more to do in terms of working out what's not going to work and working out what is going to work

Speaker C: just in that proof of concept. Right. I mean, you mentioned there about regulator, I suppose, working in the Gulf. How does the regulate. What are the challenges in terms of, I suppose, data regulator, AI in the Gulf versus, you know, working in the UK and Europe. Is it more difficult, do you think, or is this. Maybe it's across the board that this is a challenging landscape. AI is still very new. You got audit requirements, you got security requirements, especially in banking. And I mean, that's a difficult one. Is that one of the challenges, the biggest challenges, moving out of these proof

Speaker D: of concepts in a way, you know, where we are right now is no different to how regulators have always reacted.

Speaker C: Right?

Speaker D: They're always, and I don't mean this disparagingly to any regulator that's possibly listening to this. You know, they act after the event because naturally they have to. Right. So the commercial sector will innovate and they will throw these things out and they will then look for forgiveness if you like. And that's only because permission can't be sought because there's no yet direction coming out of the regulator about what you can and can't do. And there's two aspects to it I think in terms of with regards to AI, there's the how AI is being used within the commercial domain for the purposes of enhancing, you know, a client experience, the operational aspects of that particular commercial endeavor. You know, how it's been utilized internally within the firm for the good of the firm and the firm's clients. And then there is, you know, and with that it's like well okay, well what are the do's and don'ts about how are you going to implement that within your organization from an AI perspective? You know, how can you prove, how can you have traceability, how can you ensure that the decision was made for the right reasons, etc. And then the other side of that is how is AI going to change the governance relationship from the regulator, uh, to the regulated? So how is AI implementations going to change that dynamic in a banking context? For example, quarterly updates and a yearly ICAP from a capital adequacy perspective seems now quite prehistoric. There's no reason why you couldn't implement some AI solutions to provide some more real time regulation relationship between the two parties. So I think that area is very nascent. But uh, certainly my observation that that's the direction of travel. You know, in an ideal world the regulators would want to employ some of these solutions to be able to have more of a real time view of how the banks, all the, you know, not necessarily banks, any regulated environment, how they are operating as opposed to being told on a quarterly basis. So yeah, we had a problem here or on a yearly basis, this is our position. So that'd be a very interesting um, development. But you know that's, that's less advanced than even the proof of concepts around how AI is being used within the corporations themselves for the purposes of improving the corporations effectiveness and experience and services to their clients.

Speaker C: Okay, I can just ask, you know, in your experience in terms of working with AI, whereas for people, I mean, where should these organizations, where are you getting to see the most success in terms of focusing their time and investment on um, AI, improvement in operations, service operations. Is there specific areas you're seeing, you know, getting more success over other areas? Or does it, there's no impact on that business case by business case? And how do you see that moving out to the future?

Speaker D: Yeah, I think there are so Many different business cases. And that's, I think again, that's the, that's also a challenge. You know, it's like where do you want to go? You know, the use of LLMs is a great example. Right. So LLMs are out there now. Everyone's using them. Um, we're using them on a personal basis. You know, we're all sort of. And the obvious desire of employees of organizations is to use them in their work because they, you know, the efficiencies are non disputed. The problem that a lot of organizations in that particular use case are concerned is, well, how do you shore up your firm's ability to not be at the wrong end of a data loss issue because an employee said take this commercial contract and translate it into Arabic, for example. And you've just exposed, you know, your organization to all sorts of things. So you know, case by case basis and how are you going to protect yourself about that? So there's still a lot of security concerns, I would say, around, you know, the AI use cases. So there's a lot of focus in that area. We're seeing and you know, as I said, some of the automation, if you like, if that's the correct word of some of the sort of manual human tasks within, you know, banks and government entities is also another area where it's obvious there are obvious use cases for AI usage. So I don't think there's a theme per se. I think it's, you know, there's as many use cases as there are people with imagination at the moment. And you know, it's in keeping with, as I've said, we're still in that proof of concept stage. I think in a lot of cases there has been some things that uh, have percolated at the bottom. Yeah, this works. We can see operational necessity for doing it. We can see the value operationally for actually adopting that. So let's adopt it. They are dropping through the hopper and being implemented. But the speed to which these things are being adopted in production is probably the tap has not yet been turned on. The tap certainly turned on to the lab environment. There's lots of proof of concepts and proving going on, but you know, not a lot of productionization of that yet.

Speaker C: Okay, can I ask you with one last question? So.

Speaker D: Yeah, sure.

Speaker C: And I think you've kind of covered it. But do you see AI replacing all these jobs or do you see it more as a co worker? Are we far away from that? Will we ever get to that?

Speaker D: I think that's a bit of scaremongering to be honest, I can't imagine that. I think there are. There are certain tasks which AI would absolutely excel and should be implemented. You know, we've covered a few of them off today, but regulated environments, you know, brave man that says, yeah, we're going to have, you know, we're going to turn everything off, humans, I mean, and turn everything on from an AI perspective. And that will run core functions within the bank. Right. You're always going to have the human interaction and the human in the loop. Crucially important. But I can see some tasks of an operation being fulfilled by an AI capability in an AI model. An AI model, whether or not that is, I don't know. Credit decisioning, uh, for example, is another. Is another area where there's a lot of investment and investigation into how AI can help in that regard. But ultimately you're going to have to have a human in the loop because the regulator will say, well, can you justify some of these decisions? So I think, you know, the utopian idea of us all going to the beach, universal basic income, I think, is not going to happen in my lifetime. I don't think we're not going to

Speaker C: move into going out playing golf every day anytime soon.

Speaker D: I wish. I wish I could at least, I could at least then at least work on my slice, which I managed to lose actually, the day, but then I found a hook, so.

Speaker C: Oh, listen. Gives you something to work on in you.

Speaker D: It does, it does, certainly.

Speaker C: Listen, Mark, great to have you on the podcast today. Thanks again for your time. Listen, feel free to plug your company and if people want to get in touch with you, please go ahead.

Speaker D: Yeah, brilliant. I think we will put the details in the notes, won't we? But, yeah, that'd be great. Thanks again for your time. Speak to you soon.

Speaker C: Pleasure.

Speaker B: Thanks, Mark.

Speaker D: Yeah,