The leadership gap holding back enterprise AI

Jason Hemingway: [00:00:00] Welcome to In Other Words, the podcast from Phrase, where we explore the future of global business with the leaders who are making it happen. Today, we’re joined by Raf Postepski, Senior Director at Alvarez & Marsal and a leading expert in Artificial Intelligence and machine learning strategy for enterprise transformation. Raf has spent almost 20 years, perhaps a bit over, Raf, helping organizations from Fortune 500 companies to mid-market firms and helping them harness AI to drive growth, improve operational efficiency and unlock sustainable competitive advantage. He advises C-suite executives on everything from AI strategy and governance to organizational design and scaling transformative initiatives across industries and borders. Raf, it’s an absolute pleasure to have you on the show. How are you doing today?

Raf Postepski: Thank you very much. I’m well. Really appreciate the opportunity to be here.

Jason Hemingway: Oh, fantastic. Well, let’s get straight into it. You know, as I said in the intro, you’ve worked with

[00:01:00] Fortune 500 companies and mid-market firms all on AI transformation. But I guess if you could start with telling us a bit about your career journey and the path that kind of brought you from consulting into AI leadership, and the experience that shaped how you work with the executives you work with today.

Raf Postepski: I started off my career in banking and finance in Toronto, Canada. But I just, I didn’t like the structural limitations that at least banking back then gave, right? So, I moved to a company called Dun & Bradstreet, and I worked in their client management aspect. And we were looking at a lot of their new kind of digital projects and digital offerings. So that was the heyday of Hoovers coming in; they were doing some stuff around marketing automation and marketing insights. And it was really, really cool. And I kind of stumbled into this [00:02:00] e-commerce startup, type of business, something like eBay, but more for B2B. And that got me more into looking at business processes, how organizations worked, and obviously the tech side as well, with the whole e-commerce aspect of it. And I really enjoyed that. Really, really enjoyed that. Learned a lot. Met some fantastic colleagues. And then I kind of bounced around a little bit from there. But I ended up coming back to that single organization and they were quickly purchased by U.S., larger U.S. business. And that was kind of my making, so to speak. After that acquisition, the CEO went to the Chief Operating Officer, who was in charge of Europe and the rest of the world and said, “Asia’s broken, go fix it.” And then I moved into, then managing consulting found me, I didn’t find it.

Jason Hemingway: Right.

Raf Postepski: And I got hooked up with the guys at Arthur D. Little, and then a newly formed [00:03:00] digital practice, and we focused primarily on digital transformation. That’s where I kind of got the bug in terms of the stuff I’m doing. And then, fast forward, you know, 8 years at ADL, and we had an opportunity to kind of move to A&M and start their AI practice. I had the opportunity to then, and then join that and see. And I’ve been there for a year now, and it’s been very, very interesting, very stressful, very frustrating, but all the good things in between. So yeah, that’s how I landed here. So, it’s not the most conventional route to get here.

Jason Hemingway: It’s interesting, people’s journey to where they are. And I think yours is a particularly interesting one, especially in that kind of going to Asia and coming back and then, jumping into AI. So let’s talk about that, kind of, AI area then. And you know, you’ve been at Alvarez & Marsal now for a year, and you’ve been doing the AI side of things for a while. One of the observations I kind of have about AI is [00:04:00] lots of experimentation, and almost rather than it being a strategic lever, we’ve got to try and elevate it into kind of that strategic lever away from kind of just pure experimentation. Not that there’s anything bad with experimentation. But in your experience, you know, how can leaders and C-suite leaders embed AI thinking into their business strategy across all departments rather than just sort of these isolated kind of operational pilots that kind of crop up. It’s almost like that whack-a-mole thing, isn’t it? It’s like every team wants to do it, you know, you mandate it as a C-team and say, “Look, you know, we want everybody to be using it.” But unfortunately, there’s loads of experiments not really unified. So, how do you kind of think about that?

Raf Postepski: I think about it in two parts. So, I think we are now at a point where AI is where the tech function was about 15 years ago. So if you remember 15 years ago, [00:05:00] tech was always part of operations, it was always under the remit of the COO, and you know, back in the 2010s, a lot of people were carving out, probably before that, to be fair, but a lot of companies were carving out technology and having its own pillar because they saw technology as a strategic direction for the company to go in because no longer was it operational. No longer was it about end user compute. Oh, my email’s not working, let’s call tech. Oh, printers aren’t working, fax machines, if we remember what those are. So, all of a sudden, technology was an enabler for the business to move faster, to think differently, to use their assets in a more strategic and holistic way. And I think we are now at a point where we are seeing that with AI. So AI, as a consequence of it being a technology thing, traditionally has been sitting in many organizations, not all, but many organizations within the [00:06:00] tech function, within the CIO. And now I think if you really want to truly exploit what AI is capable of, you need to carve it out of technology and have its own pillar. Many organizations are moving to a Head of AI that sits outside of tech or a Chief AI Officer, and I think it’s following a very similar pattern because it’s not just bound by tech. Just like back then, technology was bound by the limitations of operations. Right now, we’re seeing, with a lot of traditional businesses, AI being limited and bound by the constructs of how technology works and how technology operates within an organization. And yes, it is a technology feature, but it shouldn’t be governed and limited by how technology is used today in businesses. So, fast forward now, what do we need to do in terms of Boards? I think, you know, number one, you need an AI strategy. You need an [00:07:00] AI strategy that informs the business strategy, but is intertwined with it. And it’s no longer good enough to say we need to be AI-powered or leveraging AI; you need to understand, okay, what does that mean for the vision, but what does that mean in terms of the strategy and the various functions that we operate? Because fundamentally, it’s going to reshape how our operating model exists and lives. So, it is a full process up and down of how you need to think about it. And too many leaders that I speak to today think you can just layer AI on top of the existing org structure or the operating model, and you don’t need to change much. And then in reality, it is a fundamental change of how we’re going to be operating. And it’s not a case of AI replacing people. And I think it’ll get to that point eventually, [00:08:00] but what we’re going to see here is the people who don’t understand how to leverage AI, who don’t know how to work with AI, who are afraid of it, they are unfortunately going to be left behind the people that aren’t and that do use it and that understand how it augments them. So, we see AI very much as an augmentation. Not pure automation.

Jason Hemingway: Yes, it’s interesting. And I think, you know, what you’re talking about really is AI as a business problem more than a kind of tech problem in itself. Because if you’re thinking you’re talking about operational change, you’re talking about a business strategy. Can you just go into a little bit more about that kind of what organizational changes need to be there? What’s non-negotiable for kind of success in your kind of opinion?

Raf Postepski: I think first and foremost, you need an overarching AI strategy that goes below the Board, so you know, N minus one, [00:09:00] N minus two, N minus three, N minus four levels of the organization need to understand what it is that we’re trying to do from an AI perspective and what that means for them and how it will change, help do things differently for their day-to-day roles. I was having a workshop last week with an organization that’s literally in mid-flow of this. And it was with a bunch of N minus three, N minus four stakeholders. And I was introduced as, oh, here’s Raf, he’s helping the organization help shape the AI vision. And they’re like, “Oh, we have one.” I’m like, “Well, yeah, we do.” Yeah, but it says X and Y that the CEOs, but we have no idea what that means. I’m like, that’s fair enough. We need to figure out the drill down what that means to you on a practical level so that you understand that, hey, you know, here’s how I can help shape this. Because everybody’s really excited about AI and they really want to jump in, but they don’t know how because that level of detail hasn’t been, kind of, distilled down. Or to be fair, the [00:10:00] organization doesn’t really know how to do it beyond the statement of ‘we need to be AI-enabled.’

Jason Hemingway: And it kind of leaves it up to people’s own, I mean, the worst case is it leaves it up to people’s own interpretation of that. And you deal with, you know, big businesses, large enterprises, as well as kind of mid-market. And in my brain, I’ve kind of got this idea that you might tell me I’m wrong, but I’ve got mid-market firms are often very much more nimble than the global, you know, enterprise. Because as a larger enterprise, you have much more complexity. So, how do you, kind of, advise them to balance, or even vice versa, the mid-market and the enterprise? How do you balance that speed of execution and agility with kind of the correct amount of governance or guidelines? Or as you said, that kind of change management idea?

Raf Postepski: Let’s start with the mid-market and the mid-tier players because they have a benefit. In my opinion, they have an advantage over many large, [00:11:00] established, call it enterprise businesses. They may not have the budget, but they have the ability to not be bound by legacy. So legacy architecture, legacy data structures, legacy processes, legacy people who’ve been there for 30, 40 years, they don’t have the burden of legacy. So, they can move a lot faster. They can try new things. They can bring in the new pieces of tech that are required to kind of help power and scale AI fairly quickly. So, they can implement. They can implement and do. What they sometimes do lack is the ability to strategically plan it for a long term. So, there’s a bit of a trade-off there. And there’s obviously the governance angle of it, of let’s just not buy everything because it’s cool and we’ve got some budget and all of us, and let’s figure out how to piece it together later. They really do need to kind of slow down a little bit and think of this holistically from an [00:12:00] architecture perspective, from a non-model perspective, to here’s the trajectory the business needs to be going and here’s how we can accelerate using these capabilities. Sometimes, they’re lacking in that thinking, so that’s where we can come in and kind of help them shape that a little bit more. On the flip side, with the enterprise guys, they have so much legacy. They don’t know where to start. They see AI coming in as like another layer of complexity within the organization. That becomes a point of exhaustion before you even begin. And people are thinking about their architecture and data incorrectly. I’m not saying it’s wrong. I’m saying it’s incorrect in terms of AI because what we’re seeing now is everyone’s thinking in terms of business systems as being the core critical thing that allows business to run, right? And where we’re moving to with AI, and I think [00:13:00] arguably we’re already there, but where we’re moving to with AI, that they’re just boxes of stuff that do things. I’m not necessarily worried about the critical business, like Salesforce CRM, right, as an example. Salesforce CRM does either the marketing lead nurturing, the sales process, and the contacts and the interactions within customers, the pipelining, all of that. I don’t care about any of those specifics. What I care about is the data that it holds. If I can extract that data and make that useful and accessible, that’s all that matters from the context of an AI practitioner. So, all of a sudden, these organizations think of these business systems as these critical things. I see them as just boxes holding data. If you can extract that data and make it more useful, make it more fluid across the organization, all of a sudden, you can move faster across most things. That’s a tech perspective. Obviously, you have a governance, behavioral and change management and a model [00:14:00] thing that you got to plow through as well, but I think the biggest critical factor for businesses, for legacy businesses, is breaking down data silos, extracting data, getting as much utility out of it as you possibly can. That doesn’t mean you put it into a warehouse or a lake or whatever. There’s better different ways to do that that give you speed and agility that you didn’t have, say, three, four years ago.

Jason Hemingway: Yeah, I think that’s interesting because speed and agility is definitely one of the things. But actually, to paraphrase you a touch and tell me if you disagree, rather than focusing on the internal data and all that, which is important, don’t get me wrong, is you’re focusing on outcome and how quickly we can get to the outcome that you want. You know, so let’s take it down a practical level. So, we talked a bit about Salesforce, but give me an example where AI, kind of, not only improved that kind of efficiency, speed and [00:15:00] agility, but created kind of things like, I don’t know, new revenue streams, new growth opportunities, anything that you’ve come across that kind of is that outcome focused for the business.

Raf Postepski: We work with a number of firms about looking at, let’s just say, the operational efficiency of it all, because that’s where a lot of people are at the moment. And there was a pharmaceutical company that we worked with that helped them kind of sort out their clinical trial feasibility, clinical trial process, right? So all of a sudden, that harmonized a lot of their data that kind of gave them a better insight in terms of how they were running their clinical trial operations, how they were recruiting patients, what the trajectory was looking, what decisions they can make on the back of that, right? So, if you have a better view of your trajectory of where you’re going, it’s not going to be 100% accurate, but it’s going to be a good chance if you’re making a good guess. All of a sudden, you can make better business decisions. And what these guys have done is they took the insights from the operational [00:16:00] efficiency that was gained and were able to then make better go-to-market strategies to get to market faster. Because in pharma, as you can imagine, it’s a first-mover advantage. I think I read somewhere a statistic, and correct me if I’m wrong, but if you’re first to market with a drug, I think you have a billion-dollar sales advantage over your competition.

Jason Hemingway: Yeah.

Raf Postepski: Or something like that. Or I think it’s four billion, but it’s a billion dollars to take it to market.

Jason Hemingway: Yeah, that’s insane. Yeah, that’s insane.

Raf Postepski: It’s something just staggering. But if you’re first, you generally win.

Speed is massively important in route to market, but also how you get there. So, it’s the decision science. It’s the decision intelligence that you instill on the back of the stuff that you do better that will get you there. So AI is very much, I look at it as two sides of the same coin. You have the operational efficiency, and you do things better with what you’ve got, but then you can make decisions. And it’s the second one that people haven’t really gone all in on yet.

Jason Hemingway: Yeah. Yeah, it’s speed with thought, I guess, isn’t it? That’s the [00:17:00] interesting part.

Raf Postepski: It’s not the machines are going to think for me. It’s the machines are going to give me the opportunity to better scenario model out what I can do without having to commit R&D dollars, right?

Jason Hemingway: Yeah.

Raf Postepski: They’re going to be able to give me the different variations of what I should be doing from an operational perspective of where I should be investing my money before I have to commit. And I think that’s the key thing. We’re speaking with a lot of customers at the moment of like, “What do I do if?” If the base rate goes up, what do I do across my product portfolio? Well, we now have the opportunity with historical data to model out to a fairly good level of accuracy of how their products should be changed. And then what is the upside, the impact of that to their customers? And then what does that mean for the margin? That’s the decision size that we’re talking about.

Jason Hemingway: Yeah, it’s interesting. In that sense, it’s about [00:18:00] monitoring and managing risk and risk in terms of almost real-time risk management, in that sense, I guess. It’s interesting. Let’s talk a little bit about, as companies, you know, you deal with global companies. And obviously, there’s lots of international different, you know, local regulations, there’s cultural differences, how do you kind of align those kinds of AI initiatives, like globally, culturally, all of those? Because, you know, we tend to talk of businesses as just this amorphous, you know, one kind of entity. And usually, you know, we think of it in terms of U.S. or, you know, English-speaking at least, but when you’ve got all those different cultural differences across a global organization, how do you sort of align? What would be your advice to kind of align those initiatives together?

Raf Postepski: Firstly, you start with what is it that you’re trying to do, right? So, I think just a broad understanding of what do we want AI to do across our organization? And then [00:19:00] you have to look at regionally, what are the different nuances that we have to cater for? Right? As an example, in the Middle East, they’re going hell for leather in terms of all things AI, but they have, you know, a governance issue with data sovereignty. So, you have to factor for that. So, it’s a case of we want to do all these wonderful things with AI, but now we have to almost ring fence the data for the Middle East to make sure it’s sovereign in the region, and the security protocols are in there that we don’t see leakage in and out. And so you have to factor for that. That’s all doable. It’s not hard, but you have to just, you can’t have a blanket. We’re doing this, guys, on a global level. Now go do it. You have to be cognizant of some of the regional differences. And also, you know, if you move back into Europe, you have the AI Act you have to deal with, right? The different risk classifications. So there’s a [00:20:00] risk angle, as you were mentioning. How do we classify what we’re doing? And then you have GDPR to layer into that. Okay, is this personally identifiable information that we have to factor in? And how do we obfuscate if we have to? And what do we do? So these are all governance and process questions and technology questions. They’re all easily solved, but you just have to understand what you’re working with. You just can’t, you can’t go like a bull in a china shop through everything and say, okay, right, we’re doing.

Jason Hemingway: It’s almost like, when we go back to your first point, isn’t it? It’s like, if it treats that as the kind of N minus three, N minus four, in the same way. It’s like you’ve got to make sure that you’ve got all of those things, people understand what they are, where those regional differences are, what the regulatory environment you’re offering in is? So, look, I’m going to lighten the mood a touch, from kind of deep diving into regulations across the globe. And we kind of had this mid-show moment, where we ask all of our guests, what’s one task, and it’d be a bit interesting for you, because you’re right in it, and I’d be surprised if it isn’t an AI task, but anyway, one [00:21:00] task, whether professional or personal, that you wish you could automate.

Raf Postepski: PowerPoint decks, I just, and Spreadsheet models. Although there’s a solve for that with the new Claude models that came out, apparently. I’m itching to try that. But, you know, there’s so much time in businesses, not just in consulting, but consulting a lot, where you have to just spend time on formulating a deck for a client or for a steerco or for a new pitch or whatever. I’m like, if you could just, if I could just tell you what to do and then you can format it and then give me all the salient talking points and just have me tweak around the edges, that would be ideal. That would save me so much time.

Jason Hemingway: Yeah, well, I knew there’d be something in there. And in true consultancy fashion, the PowerPoint was mentioned, which is good. I’m only joking. So, look, if we get back into it, you started talking about this idea of human-centric kind [00:22:00] of, and technology working together. And I’ve heard you talk before about highlighting build for advantage, buy for parity. But how should leaders apply that principle? If you could explain that principle to start with. And then, how do business leaders put that into practice, and ensure that AI adoption enhances human capabilities, rather than replacing them, particularly in sort of decision-intensive kind of roles?

Raf Postepski: Obviously, over the last 24, 36 months, we’ve had the explosion of generative AI and ChatGPT and all that. And it’s been great. And it’s also been very annoying because everybody thinks AI is GenAI, versus kind of GenAI is like 1% of what AI is. But so from that perspective, a lot of, in the early days of GenAI, everyone was like, okay, we need to build our own GenAI kind of resource to help with our knowledge management and things like that. And what we’ve seen over the last few months, probably over the last [00:23:00] year, a year and a bit, is the value is now no longer really in the model because everybody and their cousin is formulating a startup, and we’re GenAI powered, we’re doing this, we’re doing that. So, the model itself is no longer really that important. But what is important is kind of further upstream, which is your data, your orchestration, how you bake in your business processes, the logic layer, all of that that the model reads from. Because the model really is just a dumb parrot, really. It’s like, go. It’s not thinking, it’s just looking for the next most logical token to answer a particular question, right? It wants to be right. And wants to give you the best possible answer. So, where does it go look for that answer? If you give it to the internet, it’s going to give you a whole bunch of garbage. If you ring fence it with an appropriate data set, then all of a sudden, you’re getting more quality out of it. [00:24:00] So, it’s how you organize your data and how you organize what the model is reading from. That is where the value is.

Jason Hemingway: Yeah. So, in that sense, you know, if the models themselves are to a degree commoditized, I guess, the businesses need to invest in what surrounds them in order to have that true sense of differentiation, right?

Raf Postepski: Yeah, so exactly. If you want to build in, if you want to get the organization to leverage GenAI capabilities, don’t worry about the model side. You can buy that in, but that model is going to need to read from somewhere. So, sort your architecture out, sort your data out, sort how you relate certain data, how you bring in external data into that data space, so you have a much better way to understand what we’re trying to do. Similarly, with agentic AI coming in now, it’s a derivative basically off of GenAI, but how agents are going to interact. So, agents right now, they’re going to become highly commoditized. They’re going to be a race to the bottom. So, you can buy those in, but [00:25:00] you still need to have them looking at the right codified processes, you could look at the right org structures, you need to look at the right data, and the frequency of that data, and how you do it. So, all of a sudden, architecture, not that it never has been, but it becomes even more important. It’s the stuff that you build around the model, not the model itself.

Jason Hemingway: So ultimately, do you think that AI changes the way companies will define their competitive advantage themselves, moving away from kind of pure products and services to kind of this, the competitive advantage is your data that you have and the speed and agility of the decision making you make. So, the velocity of that, I guess.

Raf Postepski: I think that’s the trajectory we’re going to. How businesses get there, they’re going to be multi-speed, different times, but like I said earlier, it’s not so much about the system anymore. It’s about what the system holds. Then all of a [00:26:00] sudden it becomes, what do we do with that data? And then how do we present that both to the business and to the customer to give them a much better way to interact with it and a better experience of using it? And that’s where, you know, the front end, the interaction layer is going to become even more important.

Jason Hemingway: So, let’s talk a little bit about that kind of interaction layer or rather, the customer-facing side of things. So, how do you sort of see, and it’s a topic dear to my heart. I’ve been in personalization and marketing for 20 years, the same as you and everything else. But how do you see kind of AI transforming the way that organizations engage their customers, build trust with customers, grow globally? And what role should marketing and customer-facing teams play in that, driving that adoption from your point of view?

Raf Postepski: Yeah, well, I think this is where kind of an ethical kind of debate comes in, right? Because you’ve got, you now have AI that’s got avatars [00:27:00] that look like you and I, interacting with people in a very similar contextual way. You can figure out they’re an avatar fairly quickly because of the way they phrase certain words, or the lack of pauses and things and how they move. But, you know, that’s only going to improve. So, I think the experience layer is going to be critical in terms of how we as consumers interact with these technologies. I think much like the value comes earlier in the value chain in terms of owning your data and how you orchestrate that data and logic. Also, the experience layer is how you deal with the customers, whether it’s a pure AI play or it’s just a really, really easy way to interact with your product, with your service, or however you want to do it. But the stuff in the middle no longer matters so much. It’s the stuff on the edges of that value chain that are really going to make a difference for people. It’s how you [00:28:00] bring people closer to it in an organic, seamless way, as if you and I are talking.

Jason Hemingway: Yeah.

Raf Postepski: That is going to be the next key, or piece, to kind of gain that next competitive edge over somebody else.

Jason Hemingway: Yeah. I think it’s interesting because, you know, digital transformation, that word was bandied around for God knows how many years. But we always used to start with, and the philosophy of it is digital transformation, not for transformational sake, and you know, the tech sake. But it’s really starting with the customer and what the customer is trying to achieve. And I think what you’ve just said really chimes in with the way this should work. Anything you’re doing ultimately in the end, from a customer-facing point of view, should start and end with what is the outcome for the customer? And are we doing what we can to help that customer across their journey to get whatever they’re doing? If it’s a, you know, call to a call center agent, if it’s a web visit or, you know, if they’re going in a store or something, how do we help them [00:29:00] on their journey to achieve whatever goal that they’re trying to achieve? So, in that sense, it’s definitely putting a focus or the philosophy is putting a focus on customer outcomes rather than just speed for speed’s sake to make us more productive, you know, as a business, you know, to do it. It’s almost rebalancing is it. Do you think that kind of or ultimately it will rebalance that idea of we can be agile, but we can actually get customers what they want rather than, you know, just blasting stuff out like the old personalization thing used to do?

Raf Postepski: It’s not, because we’re all getting spammed across every single channel from everyone, every business. So, it’s going to be, I think, a quality of interaction versus the quantity of interaction. Now, interaction comes in many forms. So, how do we build in the intelligence behind that to say, you know what, I think it’s time for Raf to [00:30:00] get a piece of comms, versus just blast comms at Raf because, you know, he seems like he might be willing to buy a car or buy this or buy that. So it’s, again, I think human-centric design has always been important and will increasingly be important. Design thinking has not had its heyday. I think design thinking is as relevant as it was yesterday. But it’s a case of now, how do we bring in some of these new techniques and capabilities, you know, with agents to say, actually, the behavioral model says we should not speak, we should not communicate across these topics, but those ones over there. And the frequency in which we make that communication, because you now have, everyone’s so fed up with getting spammed, and you have a very finite amount of time to then grab somebody’s attention. So, it’s quality of interaction that’s going to [00:31:00] matter.

Jason Hemingway: I love that. I love the idea of quality of interaction. I always thought with personalization, it was just as much about not spamming and taking things out of the equation as much as it was putting in. It’s about appropriateness and relevancy for that person, that interaction, when that person has that interaction with the brand. But anyway, I could talk for hours on that subject. I won’t bore the audience too much with that. But we’re getting towards the time now, and it’s been fascinating, this discussion, so thank you. But if you could leave executives or business leaders with one actionable mindset shift to unlock AI in their organizations, what would it be?

Raf Postepski: Understand what you actually want and are asking for. Saying we need AI and we want AI. It’s okay to not know what you’re talking about. It’s okay to have a me too mindset in terms of we’re also going to do this. But ask yourself the next question, which is, do I really understand what this is? And have the humility to say, [00:32:00] no, I actually don’t. I probably need some help. Whether it’s from someone within lower ranks of the organization that gets this stuff and is super excited about it, or an external or a peer, or just go on a forum and say, okay, I need to do this. What do I need to do? But have that next level of curiosity of what the impacts bring.

Jason Hemingway: Yeah, I think that’s fantastic advice. So yeah, definitely something that you learn, isn’t it? It’s moving so quickly. I don’t think you can stay on top of it so easily without having that curious mindset, right?

Raf Postepski: By the time we land one thing, there’s going to be four other things that are going to either make that obsolete or are going to completely change how you think about that one thing you landed.

Jason Hemingway: Yeah, accept the unknown and embrace it, I guess. So, talking of the unknown, if we can look like a few years ahead, what AI capabilities do you think, and I won’t hold you to this, although, you know, maybe we can chat in a couple of years’ time and see if you’re right, [00:33:00] what AI capabilities do you expect most will reshape how companies expand globally or compete at scale? What are you sort of thinking currently today?

Raf Postepski: I think in the next, 3 to 5 years is a long haul, let’s shrink that down to the next 18 to 24 months. I think companies have started on the AI train, and they’re all moving at different speeds, but they’re all looking at the efficiency gains. They’re all looking at how do I make my operation leaner? How do I do more with less or more with the same amount of people? Or how do I augment things here and there? I don’t think many organizations, I’m not saying all, but many organizations haven’t quite figured out how do I make better decisions. How am I able to harness all these new things, this new data landscape that we’ve had to reconstruct because of what the models require? How do I then harness what I have in front of me to make better decisions, not just kind of on an operational scale, [00:34:00] but on a strategic scale? So, how do I know which organizations I should acquire or which markets I should enter? Or do I need to sunset some products just because, whilst the metrics and the OKR seem okay, I don’t see it as a long-term kind of cash cow for us. What else do we do? Looking at their business, their markets, their competitors, and making those decisions is going to be critical, and the business that win are the ones that do this quickly. And what I’m referring to is causal machine learning. It takes the correlation out of machine learning and focuses on cause and effect, and gives you the ability to scenario model grand ideas of what if something happens across the base rate, and I adjust my product set, and my competitors do the same thing. What are the things in the margins that I can do that they won’t be [00:35:00] thinking about, that’s going to give me that edge?

Jason Hemingway: So, for them to start thinking about that, what’s a first step, just a practical bit of advice to do that? What would you suggest to a business leader? Where would you start?

Raf Postepski: I think you need to start from what you’re trying to do from an AI perspective as an organization. Understand really what you’re doing and what you’re trying to do. And then ask yourself the question, what else can I do with this? Because oftentimes people don’t know. And it’s the unknown unknowns and the inability to ask those questions that kind of hinders us. And be fair, I’m just as bad as everybody.

Jason Hemingway: Yeah. And you said that sort of starting point is to get that strategy sort of documented, start thinking about that, put it down on paper, perhaps even start.

Raf Postepski: Yeah. That strategy that then flows into a roadmap of sort that has key transitions built out across a timeline, and [00:36:00] say, you know, we’re going to have this plan in place, it’s not going to be locked down, it’s going to be directional, and we need to have the understanding that things are going to shift and change based on what happens. But that’s okay. That’s fine. We’re just going to roll with it.

Jason Hemingway: Brilliant. Okay. Look, we’re now into the very last stretch with our quick-fire round, which we ask everyone. So, I’ll give you a couple of questions and just shoot from the hip, see where you end up. So, a company you admire right now for AI adoption, and why?

Raf Postepski: I’d say this is probably a cop-out answer, but I’m going to say Netflix.

Jason Hemingway: Okay.

Raf Postepski: So Netflix, Netflix is awesome for two reasons. They house and sit on an exorbitant amount of data. And what they’re then able to do with that data is they can release relevant content to [00:37:00] people like you and I, who have very different preferences, very quickly. I think the production time from idea to finished product at Netflix is so much faster than the big studios. Why is that? Because they’re sitting on data. They understand their user behavior. They understand the user preferences. They understand what tracks better because they’re A/B testing constantly. And all of a sudden, they’re gaining that data. And then they’re using it, making decisions for content, making decisions for people. And then also the ability to hyper-personalize across a vast amount of, I don’t know how many millions and millions of users they have, but what you see for your content on Netflix to what I see could be two different things, even though we’re both middle-aged white guys. But we may have different preferences, and their ability to curate content to us individually is amazing.

Jason Hemingway: Yeah. It’s interesting because Netflix is a fantastic business and has grown kind of from the digital era, [00:38:00] right? So, they’re almost like their very existence is born in competitive advantage because they’re built for this kind of thing. So, it’s interesting, but great, great, great company. So, growth in one word. Global growth in one word.

Raf Postepski: Curiosity.

Jason Hemingway: Good one. And final question, and I’ll let you go. One book every leader that’s scaling AI, what should they read?

Raf Postepski: Good question. I’m literally about to press buy on a book called The Competing in the Age of AI, Strategy and Leadership.

Jason Hemingway: Competing…

Raf Postepski: When Algorithms and Networks Rule. It’s written by a pair at Harvard, Karim Lakhani and Marco Iansiti. But it looks at alt-model transformation, looks at network effects within organizations and data, looks at the strategic blueprint, what do we need to do? [00:39:00] How do we need to do it? Where do we build flex? Where do we design for uncertainty? All of those really cool things. And I’m looking forward to get stuck into that one. But there’s also, it’s an oldie but a goodie, The Lean Startup by Eric Ries. It’s not an AI book, but it’s a case of how do you do stuff fast and design for agility. And that’s still valid for AI and always will be.

Jason Hemingway: Excellent. Well, there’s two books for the price of one there. Well, maybe not the price of one. You’ll probably have to pay for them both. But look, Raf, brilliant, fantastic discussion. Thank you so much for joining me on today’s session and telling us all about AI and what you’re doing and where it’s going, and the thoughts, and more importantly, I think, to give people some advice of what to keep thinking about, because like you said, it’s changing so quickly and you need to just get a hold of it, embrace it and work out ways that your business can use it to its advantage. So once again, thank you very much for [00:40:00] joining us, and I’ll let you go. Raf, that was great.

Raf Postepski: Thank you, Jason. That was great.

Jason Hemingway: Well, that’s it for another episode of In Other Words, a podcast from Phrase. I’ve been your host, Jason Hemingway, and a huge thank you to Raf Postepski for joining us today. If you enjoyed today’s episode, be sure to subscribe on Spotify, Apple Podcasts or your favorite podcast platform. You can also find more conversations on leadership, growth and what it really takes to scale globally at Phrase.com. Thanks for listening, and I’ll see you next time.