Aris Valtazanos: Start with the Pain Point, AI Isn't Always the Answer

Speaker A: Don't just try to solve everything at once, but you know, find the one or two or three workflows to begin with where you feel this can add more value. Think of like, know what the really big pain points are in your business and then think if AI, uh, is is the right solution for that specific problem. So it's not just. No, you hire one person and it's all magic. No, it still is a coordinated effort from different parties that is required. Foreign.

Speaker B: Welcome back to the AI Pathfinder for Private Equity podcast. AI uh Pathfinder helps private equity firms make sense of AI and make it work. It's an expert network built on insight, experience and connection, bringing the right people and ideas together to turn AI from something firms are curious about into something that delivers real results. If you'd like to attend one of my regular AI strategy briefings in London, Manchester and soon New York, please check out the show notes for details. This is our next episode in our, uh, AI uh Operator series, Conversations with the people doing the hands on work of leading AI adoption inside private equity firms. What's striking as I record more of these conversations is how different the paths into this role are. Ah. And how that might shape the way people approach it. Today's guest is a case in point. Aris Baltazanos is Head of Data and Analytics at Oakley Capital, one of the UK's most established mid market PE firms. Um, known for backing ambitious founder led businesses in tech, education and consumer. Aris has a PhD in robotics from the University of Edinburgh, spent nearly six years at Quantum Black and a stint as Chief Scientist at Woven Light before moving into pe. That combination of deep technical foundation, large scale consulting experience and senior scientific leadership gives him a lens on AI and PE that will be well worth exploring. Aris, welcome to the podcast.

Speaker A: Thank you for inviting me Steve. Great to be here.

Speaker B: Really looking forward to the conversation. Well, let's try and understand a little bit more of your background here and your route into pe, because it's probably a little bit different from perhaps uh, some of your peers. What point did private equity come onto your radar and what made Oakley the right move for you?

Speaker A: Absolutely, yes. Let me roll back a few years perhaps and start with when I started working, uh, in AI. My first exposure in AI was perhaps a little bit earlier than most back at my undergrad days, which was nearly 20 years ago. But at the time, uh, where I studied at the University of Edinburgh was I think the only university at the time, uh, offering an undergraduate AI degree. So that sort of planted the seed if you like, at the time where AI perhaps wasn't top of uh, everyone's mind. And that led to me doing a PhD in robotics where again I got an exposure both to the more academic side of AI but also the more hands on building things and solutions. My PhD involved lots of experimentation with physical uh, robots. And at the time I realized that I really enjoyed doing hands on work solving real world problems. Sometimes I even like the messiness and the challenge that a lot of these uh, problems entail. And that's led me to a career in industry in the first instance where after my uh, Ph.D. i joined at the time a little well known startup called Quantum Black which had spun off um, the Formula One. So all the founders of Quantum Black initially worked at Formula One teams looking to find efficiencies in the car manufacturing process and then eventually scale this approach to other sectors like banking, uh, pharmaceuticals, advanced industries and many others. As you may know and as a lot of people may know, Quantum uh Black was eventually acquired by McKinsey. So then we went from being a really small startup of about 20 people to being part of a much larger organization but still focusing on working with clients and helping them understand how data can solve a lot of their real business problems. And then after having spent a few years at Quantum Black, um, working with larger enterprise clients, um, I got my first exposure to PE as part of working at Woven Light where we applied some of the similar learnings and approaches we had at Quantum Black, but more from a PE lens and focused on working with uh, PE owned portfolio companies. Um, that was if you like, not a segue to me joining um, PE firm like Oakland being then more the in house resource supporting um, a larger portfolio of portfolio companies but then with a similar lens of trying to identify where data analytics AI can uh, create value, um, and then helping them find the right approach. So that was a little bit my trajectory. So more academic version of AI to more generalist consulting, then more to PE focused consulting and services and now more to being an in house PE resource.

Speaker B: Yeah, thank you for that background Aris. Do you feel that the private equity has that sense of urgency? I realize we've gone through a bit of a uh, curiosity phase to something that's perhaps a little bit more I suppose reflective in terms of where you've been with AI, which is probably more scaling, do you feel that there's an alignment now perhaps more so with your peer group that we are now beyond that to actually some scaling. And another question actually is are you also looking at the firm's adoption of AI as well as the portfolio.

Speaker A: Yes, yes to all of the above I think is the short answer. So I think no, uh, PE firms have started looking at data analytics and AI in various guises I think over the past few years. But I think you're right. It's really in the last two, three years with everything happening around generative AI, that's perhaps some of the sense of urgency has increased and indeed the level of resource that a lot of PE firms um, are deploying. I think the approach can vary a lot by private equity funds because uh, it also depends on the nature of the investment. So how a firm like Oakley, which is investing in lots of different sectors with companies with very different characteristics to each other, may be very different to another RP funds that are much more thematically concentrated where perhaps they can take one approach and scale it more easily across the portfolio because all of their companies have similar uh, characteristics. Um, so in short, I think you need to adapt your approach to the characteristics, um, of your fund, of your investments. Um, also I think the broader approach, um, funds have around value creation and how they like to work, um, with our portfolio companies is important. So at op, for example, we don't have a really big portfolio team. Like we focus on having experts covering specific areas and then we like to bring in external help where possible to uh, support our portfolio companies. So again, you have to be very mindful of these things because these really inform how you approach and support your portfolio companies. To your second question. Yes, the internal side is just as important as how you support your portfolio companies. I sometimes like to take note that the internal fund side sometimes is like just a very large portfolio company in its own right with again different set of stakeholders, business units trying to identify where uh, AI can create value and trying to find the right solutions and approaches very much like our portfolio companies do. And equally like some of the underlying challenges of like how you best integrate data, how you find the right problems, what's the longer term outlook from a portfolio perspective? Like, there's lots of common themes that apply on both sides of the world.

Speaker B: Yeah, sometimes the P firm, um, acts as a bit of a sandbox for opportunities that might be able to uh, be a cookie cutter across the portfolio. Just coming back to your role that you've got and obviously the background as well. Something I'm seeing is that the AI operator, and I'm putting you in that bucket, Aris, is that it's something that's growing within PE firms and that's Right across the lower, mid, mid and larger cap firms. Um, I'll be just good to get your perspective on that role and how that's evolving really within private equity.

Speaker A: Definitely. So I think the role again as I mentioned earlier, it can express itself in different ways depending on the focus of the funds and the size of the teams they have. More broadly to begin with, I see it very much as an enabling role, especially when you have a small number of resources in house and a larger portfolio of companies to deal with. First of all, you know, you need to be able to prioritize and say this is a company or an opportunity worth exploring further or this is perhaps something that we should revisit uh, later. So being able to decide where to spend your time, where are the higher value problems, um, and ultimately where you find a lot of synergies between problems that can apply and solutions that can apply across different companies is something that's, that's really important to help you scale uh, your own. I think. No, there's always a risk sometimes that you can get dragged down too deeply in one company and perhaps lose sight of all, of all the rest. So for me that's always the, the keys like finding this balance between the breadth of the portfolio or, or the fraction of the portfolio you cover and how deeply you can go in each.

Speaker B: Uh, yeah. How do you avoid you yourself becoming a bottleneck?

Speaker A: It's, it's a good question. I think there's, there's a few ways. One is like when, when I decide where I will spend my time a bit more deeply, I always try to find more, smaller, more self contained problems if you like, where there's no risk of me um, getting dragged in and spending a long time as a settling side ah of the rest of the portfolio. So finding like smaller self contained problems that you can then easily scale and repeat across the portfolio is always a good starting point where you need to decide where you want to focus a lot of your time. Um, then equally finding other examples of use cases where you can in some cases find good vendors or partners you can work with and then you can scale them more easily because it's easier to say we've done this in one or two companies now a few more can do it I think. No, portfolio companies always like relating to uh, what other companies have, have done and found successful. So, so we always try to draw these parallels and show problems that have worked well in other places in the portfolio.

Speaker B: How much you spendings on the hands on practical side because it, it's, it's impossible not to want to use the tools and the fact it's becoming easier, you can do what you could have done in weeks now, in hours and days. How much is that taking your time and how much actually do you find that the most interesting part of your role?

Speaker A: Uh, honestly it's very mixed. I would say you know there's probably a good 20, 30% of my time in any given week I uh, spend on more hands on work or helping companies with specific problems. The rest is more trying to enable broader solutions to be built. But I think you're right, it's a, it is a very delicate balance. But sometimes you need to almost lead by example and almost plant the seed if you like by, by building something small and then getting the excitement for, for building something bigger. As you well know, like sometimes getting started is the hardest part in the journey for a lot of PE portfolio companies.

Speaker B: Let's talk a little bit more if we can about Oakley and I'm not asking you to share anything commercially sensitive here but where have you seen AI change how something gets done either at a firm um, level or across the portfolio? Not the pitch, the reality.

Speaker A: Um, yeah. So if we start on the portfolio side I think probably there's two broad families of opportunities and use cases. I think there's some more cross cutting use cases and areas where we hear a lot about like marketing software development. These can apply in um, a lot of companies of different characteristics and sectors. And then you have like more bespoke if you like solutions and use cases that only work um, on some of your sectors or companies but then can potentially go a lot deeper. Um, and if I'm honest, I think it's the latter category where so far I've seen create the most value at least for some of our companies. So, so think of some examples. We had um, legal tech business that we exited uh last year which started as, as a, as a data data business, uh, providing a large corpus of legal case to lots of law firms and other legal clients across the world and, and for them, no they had spent many years curating their data having in a really good shape. So when AI came in it was made perfect sense for them uh, to add like an AI product on top of their data product and that really helps scale their business and make it even more uh, more successful. Similarly some of uh, our education businesses have a focus on building really sophisticated AI tutors and assistants that are part of their core education offering and ultimately lead to a better student experience. So some of these examples I Think have been really good successes and really good examples of how AI works well in practice. When you, when you already have a good found, you have a good business, uh, problem, you're in the right market or segment and then AI fits nicely into that narrative and helps you really augment your offering to the clients. However, unfortunately it's quite hard to generalize some of these solutions to the entire portfolio because a lot of them are very specific to specific sectors and use cases. I think if you look at this other bucket of the more cross cutting themes like marketing software development, I think there's lots of good early evidence, especially on the software development side, that AI can lead to huge efficiencies and time savings. Not only can lead to organizational models being redesigned a lot uh, around AI, but I think we're still at the early stages and I think we're still trying to fully quantify what this will look like and what does this actually mean for businesses and the customers and the products that are being delivered.

Speaker B: Has anything changed over the last few months with Claud is making waves, isn't it?

Speaker A: Yes.

Speaker B: Um, so has the sentiment or the urgency changed around that? You talked about business models there.

Speaker A: Yes, I think there has been definitely a change in the last few months because you know, in this specific example, I think if up until the end of last year roughly there were still more mixed feelings around AI coding agents. For example, like some people said they found them helpful, but there were also a few people not too happy with the quality of the out. I think with some of the latest releases, as you mentioned, the needle is moving more towards like, yeah, this is something that's working or is clearly going to work in the very near future. So how can we embed it um, more deeply in our organizations? Having said that, I still have seen a mixed level of reception by different companies. I think some are fully on top of it now from day one. They really focus on integrating it, embedding it and their software development process. I think especially some of our technology companies with smaller teams of engineers think like 2010, uh, 2030 engineers, where it's often easier to embed some of these tools than if you had a much larger software team. There's still some cultural challenges in a few others that are perhaps haven't been as quick to adopt it fully, see the benefit yet, or are just hesitant to start with it. I'd say overwhelmingly most companies are getting on top of it, but we still see some level of resistance and uncertainty uh, with a few others in the

Speaker B: many conversations I Had the aim is to obviously support your portfolio as much as possible but don't sort of mandate. But it's getting to such a point now where there needs to be a little bit more persuasion maybe is the right word. I don't know. Do you agree?

Speaker A: Definitely. And it's a delicate balance because portfolio companies need to own this at the end of the day because if it's overly mandated or enforced then they will never really own it or they might still rely on external support. I think you need to get to the point where you convince everyone of um, the benefits in the way that you don't have to force it on them. They ask it for this themselves. And I think it's a bit of a mix of a more top down and a bottom up approach that meets somewhere in the middle. Clearly where we've seen it work there's been strong vision and adoption from the top. So if you don't have someone in the C suite who really will be the advocate and the supporter for it, then it's really hard for this to permeate down the organization equally. If you don't have the champions on the ground who will actually go and build things with these tools and show how this works, um, this will never materialize into something more concrete. The ones that I've seen make this work, there is this level of alignment across the different dimensions.

Speaker B: Let's take a look at from the inward sort P firm uh, and how they're adopting. And as you know I run a benchmark study quite regularly now across the peer group uh that we have come together and a particular role, the investment and diligence um role is rating AI maturity lower than any other group. So I think it was. 84% believe their firm is still exploring or piloting. Arguably these people are at the heart of what PE firms do and could gain quite a bit from AI assistance. What do you think that might be? And do you see there's a real untapped opportunity here?

Speaker A: I think there definitely is. And the reason why um, a lot of or nearly all PE firms are exploring this topic is that there is a lot of this work that fits nicely with the capabilities of AI systems. So going through large volumes of data to extract insights, generating documents, the documents in a more automated way, generating deeper insights and sort of triangulating across uh, different data sources. All of these P and especially investment related use cases fits very nicely with um, with what these AI systems can do. I think though why you mentioned that there, there have been some, some challenges that this this is often harder to do in practice as you know with all of these tools it's, you get lots of really nice looking demos that show you, oh yeah, you can generate an icy memo in a matter of five minutes and it all looks very nice and polished. But again when you try to integrate it with your systems then you realize that not all data, uh, is as well structured or easily accessible to a lot of these models. Oftentimes the really valuable data needs to be controlled in a very specific way so that only people who should have access to the data, uh, have access to data and it doesn't leak to anyone else. So there's, there's lots of, of challenges and constraints that come in into working with, making these tools work in practice. And I think the second point, which also links to what we were talking about, the portfolio earlier, there's, there's also the cultural side of it, so how you convince people to change their workflows. And again, because many times a lot of these tools come with a lot of hype and promise. Like a non technical user might spend a few minutes or a few hours working with it, if it's, if it's not giving them the answer they expected to get, then they might be quicker to abandon it or revert to their old ways of working. So how you introduce these tools at the right level of promise and expectation to make people adopt them, not just uh, abandon them when they don't give the answers they expect them to get. I think that's also a very critical part of just getting the technology right.

Speaker B: Obviously a lot of focus is on reducing the amount of grunt work so you're able to, to do more with, well with potentially less. How do we go from that to actually affirm using AI to differentiate itself?

Speaker A: It's uh, it's, it's a, it's a very good, very good question and I think it touches on some, some of the points I mentioned earlier is that you need to start by, by breaking down um, the problem into, to the right sub problems because again the AI investment process is quite complicated and there are lots of different sub activities that you go across depending on the deal stage, the uh, deal cycle, depending on your role as well, whether you're in the investment team or IR or some other operation role. So I think where you need to start is like don't just try to solve everything at once but you know, find the one or two or three workflows to begin with where you feel this can add more value and that's typically a Combination of where people spending a lot of time doing manual work that can be more easily automated, where there are like some more structured templates or tasks that again are uh, repeated time and time again. And I think that by starting in this more focus way then you can build on it and start seeing some of these solutions apply more broadly across the investment life cycle. Sometimes there is a temptation to solve everything at once and that's where usually things aren't as successful.

Speaker B: You've mentioned the challenges sometimes why things might move a bit slower. And the obvious one is around data readiness, data foundations. But when we asked in the um, so earlier this month around just agentic, there seems to be sort of perfect distribution across four areas really. Governors, integration into systems and workflows, talent and capability, data quality, nothing really dominated. Perhaps you know, pulling on from your experience maybe at Quantum Black and working on AI implementation across the industries that you worked on there, does that spread surprise you or is that what you'd expect?

Speaker A: Not really. I think perhaps people weren't sure which one to pick and they kind of randomly scratched the four. But I think it's a fair reflection and in some ways when I look at what's happened in the last three, four years, like with generative AI, I think there's definitely what's possible with technology has definitely gone up and the potential of what you can achieve is definitely greater. But equally a lot of these other factors have stayed the same as you said. So if anything the gap of what's possible and where you currently are has grown in some respects. And a lot of these topics around talent, data quality, governance, they still remain to be addressed and still are as important regardless of whether it's generative AI or what was there before. Um, generative AI and I think especially in the PE world there's a lot of businesses that, that haven't spent a lot of time and resource in the past in a lot of these topics. A lot of these businesses have grown very fast or there's lots of buy in builds where you have lots of different organizations, smaller organizations coming together to form uh, a larger one. So clearly there is a lot of legwork that needs to be done around data quality governance before companies can start really seeing some of the benefit of these more, more uh, advanced solution. I think I would, I would add one more to this list. Is that also finding the right uh, the right problem to, to solve with AI and making sure that AI is the right solution for that, that problem is another challenge I see a lot because I think though With a lot of the, the talk now around AI, people really feel they need to start with okay, AI is a big thing, we need to find somewhere to, to use it. And whereas I think you need to almost go in the opposite direction, think of like what the really big pain points are in your business and then think if AI is the right solution for that specific problem or if there are like some other uh, other possible solutions or other things you need to address, um, first. But sometimes I agree you need to have a more principled and structured way of thinking of all the other factors and not just the specific technical solution that you're implementing.

Speaker B: Yeah, and it depends where you are in the hold period as well, doesn't it? You know, because when does it go from you know, action to maybe creating differentiating products practically doing something to just providing the foundations that will support the next cycle.

Speaker A: Yes. Uh, and I think a lot of this often is under appreciate or underestimated because let's face it, a lot of isn't just exciting to talk about, you know, getting your, your data in order or governance. A lot of this is often overlooked but it's uh, it really underpins um, and determines that access a lot of the, the AI initiatives that sit on top of it.

Speaker B: So just looking ahead a little bit here, and I know that's really hard to do when things are changing so much, but what are you focused on right now and, and perhaps for those with similar roles or about to take a role within private equity, what, what should they be paying closer attention to?

Speaker A: Yes, I think it's um, split into two parts. I think as I mentioned earlier, a lot of the underlying topics around data quality, governance, talent, the right infrastructure, like I think a lot of these topics will still stay as important as the, as they are like especially for some the portfolio companies still earlier on in the journey. So as I mentioned earlier, yes, there's uh, maybe we'll have many more iterations and different uh, of models and solutions coming around but I think a lot of companies will still need to focus on getting the fundamentals right. The same way they should always would have to be doing. I think what's changing from my perspective and the role that I'm doing is like getting better awareness of specific solutions and use cases that can create more value and how that landscape is changing. And now uh, uh, with a lot of the tools that are being developed. So looking at what's going now with software development, some of the implications that AI has on marketing or on product development for some of our companies or equally on back office automation, which is another topic that is a lot of companies are considering about. These are m much faster moving spaces. So understanding what kinds of solutions work well, which ones don't work well, what are good starting points, uh, and solutions that companies can get going with without necessarily spending too much time on anything else. I think that is something I expect to evolve and something that I, and I expect others in my role will keep a close eye out on for the coming year. And to your second question around what should someone moving into a role like this should consider about? No, I think it's a role that requires being very adaptable. And especially when you have to deal with a large portfolio of companies with different characteristics, you always need to keep an open mind, always need to be flexible and think of different ways of supporting your companies because some of them might have the resource and not need that extra bit of advice, um, and a nudge in the right direction. Others are starting with a lot less and need to get, need a different kind of effort to get going on their AI journey. So it's important to be flexible, keep an open mind, always listen to your peers because uh, they often go through similar problems and indeed a lot of the important insights I've had are from speaking to peers in similar roles and hearing what's working for them, uh, and what's not. And I think, you know, just, just embrace the fact that it's a very fast, fast moving world. A lot of us always feel left behind, that we wake up in the morning and there's something new that's happened overnight and we always feel like oh, we're not up to date and we missed out on everything. But at the same time a lot of what, what we do still remains valuable and, and we should still focus on, on doing a lot of the things that we're already doing around supporting our companies to, to get going.

Speaker B: The final question and, and uh, and perhaps we could just flip that last one. Advice for a PE firm, um, that's looking to hire an AI operator, what advice would you give them to avoid them setting up to, to fail?

Speaker A: Yes, it's a, it's a good question. Assuming this will be their, their first hire in this space. I think, you know, it's important to, to find people with a good mix of skills because I think these AI operator like roles require a good mix of technical and business skills. I don't think you necessarily need to be the most technical person or the, have the strongest consulting toolkit. But having a good mix of both will enable you to build the right level of rapport with portfolio companies, help them turn the opportunity to something more tangible and not just something more abstract. So I think, you know, look, look for people with an intersection of these different uh, skill sets is important because I think they both really matter, um, in this kind of role. The other piece of advice is like to be prepared that this is a, ah, longer term journey. So a lot of these topics are genuinely hard to get right and practice from the moment you bring in someone through the door, um, till they get familiar with uh, the portfolio. Find the right contact points within each company, find the right problems to tackle and then ultimately get them going on. Some of these initiatives, like a lot of these efforts take, take uh, time and also alignment from the rest of the business. Because the AI Operator is one part of the solution. Uh, there needs to be strong alignment from the deal teams, the management teams, whichever external vendors or partners you work with. It requires a lot of different parties to coordinate to make this a success. So it's not just you hire one person and it's all magic. No, it's, it still is a coordinated effort from different parties that is required.

Speaker B: Yeah, I think the last thing you want to do is isolate this as sort of. Yeah, needs to be right in the middle there, doesn't it? Between, between everyone. Aris, that's been a brilliant conversation, really enjoyed it. Thank you so much.

Speaker A: Thank you very much for inviting me, Steve.

Speaker B: For listeners who'd like to follow Aris's work and find out more about Ukulele Capital, we'll include the links in the show notes, but that's it for the latest episode of the AI Operator series. If you're in a similar role, would like to take part in future conversation or join our growing AI Operator, please get in touch details in the show notes, but thank you for listening and we'll see you next time.