Sept. 9, 2024

Shaun McGirr: practical considerations for Private Equity firms

Shaun McGirr: practical considerations for Private Equity firms
Shaun McGirr: practical considerations for Private Equity firms
AI Pathfinder for Private Equity Podcast
Shaun McGirr: practical considerations for Private Equity firms

In this podcast episode, Steve Budd interviews Shaun, the Chief AI and Data Officer at DevOn, a software development company. They discuss the intersection of AI and private equity, the importance of understanding AI capabilities, and the potential for AI to unlock value in portfolio companies. Shaun emphasises the need for a practical and iterative approach to AI adoption, starting with small efficiency gains and gradually scaling up. He also highlights the importance of blurring boundaries and rethinking processes in order to fully leverage the potential of AI.

AI Pathfinder Private Equity AI Strategy Meet-ups:

https://www.aipathfinder.co/private-equity-ai-events

Have a question or would like to suggest improvements, please contact steve@aipathfinder.co

SPEAKER_01

Well, hello and welcome to the AI Pathfinder for Private Equity podcast. Bit of a mouthful, but we're sticking with that uh name for now. I'm in the process of launching a new expert network focused on AI in private equity. With any new idea or business, one of the most important things and interesting things for me is to speak to lots of of people that this service may affect. And I've been doing a lot of that lately, and and every conversation I'm learning so much. But I'm not sharing that. It's all for me, which is great. But I think that a lot of the conversations I'm having other people could could be hearing and learning from. Hence the uh the launch of the podcast. And over the coming months, I'm gonna be talking to those that are very much in the private equity space and those that are on the value chain, AI value chain. And one of those is here today to talk to me. And I'm really looking forward to to speaking with a chief AI officer, no less. And I definitely, if we look at his profile, he can he definitely deserves that that job title. So welcome, Sean, C A I O at Devon. Um thank you for joining me.

SPEAKER_00

Thank you. Glad to be on the AI AI F P P P P E podcast. There's another B in there.

SPEAKER_01

Yeah, there you go. You've you've helped with an with the acronym. Maybe maybe that that'll just spin off the tongue after a few of these episodes.

SPEAKER_00

Yeah, maybe, maybe. Yeah.

SPEAKER_01

Name it. Let's kick off. Let's kick off and just want to learn a little bit about you and your company. Obviously, you know, as much as we can in the context to our listeners, investors, and their portfolio company. Just give us a bit of background for us, please, Sean.

SPEAKER_00

Yeah, so I work for Devon, uh, which is part of a larger group of companies, actually. So the group goes about back about uh 18-20 years, and it was founded originally as a transformation consultancy around agile decision-making ways of um working, helping humans work together with other humans in a different way. And that launched in the Netherlands, and it had lots of um Dutch organizations as as customers, um, some of them large enterprises, some public sector organizations, and then started to get into a little bit of helping private equity companies with their recently invested companies change the ways of working of the of the software companies, especially that they had uh purchased, and so they had inherited some ways of working that maybe weren't optimal or modern or or fast or agile. And then eight or ten years ago, um the software development that we started to do off the back of that was split split off, and so the the uh the the parent group is called Vada, uh it's uh and the principal investor is uh Vikram Kapoor. Um, and so he has sort of a stable of companies, uh, one of which is Devon, which grew out of our sister company ProAwareness. Um and so ProAwareness continues to help organizations change their entire organizational setup and ways of working, and then we at DevON focus on the software development lifecycle aspects of that. Uh, and we uh now have um about 400 uh uh consultants and developers who can also augment the work of those teams and accelerate that journey to what we call we call high performance. So we we're we're privately owned uh and we're sort of in it for the long run across that value chain of those um companies, uh, and we are expanding into the UK as well. So that's why me, as someone based in London, is the chief AI and data uh officer when the company looked to okay, how do we expand into the UK? Uh starting that thinking a couple of years ago, it's also when the company started thinking about AI. Uh, and it's a natural and hot topic. You've it's just easier to ride the latest wave when you're taking what you do in one market to another market. With something as powerful as AI, a general purpose technology, it it really allows me to burn away all of the other stuff that has accumulated over 10, 20, 30 years about data and technology and decision making, and actually get to the essence of what I care about, which is how do we help humans and the organizations they inhabit make better decisions. And this is that moment for me. So that's why I'm in the game and uh all in on AI. Not because I don't care about data anymore, but uh data is the most important ingredient in AI, but it allows me to get to a business problem faster. Half the time, maybe more than half the time, that business problem. Once you use AI to get people to talk about it, you realize and they realise no AI is needed to solve this problem. And that's a great value outcome from AI, realizing we don't need it. All we needed to do was talk to each other, write down the steps, and automate that process.

SPEAKER_01

Thank you, Sean. That because it's AI is now in in everyone's hands, is that that could become for someone that's that's got your background and and discipline and and focus, is that you know, does that become then a bit of a burden because everyone's talking about it and and not many people understand it and not just understand it but how it can really solve problems? Does that yeah, is that causing a delay to actually getting the value out of it for you?

SPEAKER_00

I mean, inevitably, yes, like at the start of any application of a new general purpose technology, yeah, that's just where we are. Probably one year ago, I would have said it really is a burden, and like everyone is just saying, but I don't know the use cases for AI. Right. And tell me how you what your job is, what do you do all day? What data do you work with? What decisions are you trying to make? So it is a it is a burden, um, but I think it's the useful kind of burden, right, that forces us to do a lot of the work that other waves of technological driven transformation did not. So the classic cloud anti-pattern was here's my big monolith application. Someone told me it'll be cheaper in the cloud, I put it in the cloud and it's slower and more expensive. Or in big data land. Someone told me I installed this big data cluster, and now I'll have insights delivered to me somehow, but actually I've got the same dashboards and the same Excels being emailed to me because uh it was too easy to just do the tech of those other things, and so the benefit of it being in everyone's hands at some level, there's much more benefit for uh using the burden for something good, I find. And the way that has become very concrete to me, uh, I've just come back from a bunch of traveling, doing training lots of different audiences and how to use AI in lots of different job roles, both within the software development lifecycle and sort of adjacent. The kind of uh the magical realization I had is that the best way you can teach people how to think about AI for their job is to actually have them talk with a large language model about that. Right? So instead of thinking too much about whiteboards and installing Python and and that stuff, you can get a hell of a long way to simulating someone's problem, simulating the data about that problem, and simulating an AI solution to that problem just talking away to AI. Um, so that's that's how I sort of use the fact it's a burden against the people who want the easy answer. At the start of every training, when I ask what people want from AI, again, half the things you could have had that from technology before AI. Why do you think AI will solve that one? Uh and then a small fraction are like, you know, I want dancing robots. Um, but most people that I'm finding now have quite reasonable expectations, and then by making them explain their ideas for AI to an AI product, they learn the limitations of AI in general without overly focusing on their use case, and then they but then they might also solve their use case realizing I don't need a whole system, I just need three the right three prompts. And now this thing that used to take me an hour can take me three minutes if I just find the right three prompts. So there's a huge creative moment that can happen in everyone's job. Disappointing thing about the burden is when and I understand why, it's it's justified, people switch off their creativity and expect AI to give them the answers without any good questions. But as soon as people make that switch, they can they can get value immediately.

SPEAKER_01

Yeah, yeah, yeah. Well let's move into a bit more in the context of private equity. Why should they be why should they be excited by by the recent um improvements with with AI? Not just excited, maybe why should they be slightly worried? And do you think there's an urgency, should there be an urgency for them to act or or should it be more of a let's wait and see type of approach?

SPEAKER_00

Depends what their investment targets and thesis are. So I think we already know some things about which industries are not gonna fundamentally change or not change as much or change slower. But some of those things are a little bit unexpected, and it does go back to what you said about you know, until you get to that kind of task level, job to be done level in your analysis of how a workforce can change or a company can change, you don't really know anything. Investors of all stripes probably like to think about how you know whole humans can be replaced or removed or upskilled or moved somewhere else or recycled into another one of their portfolio companies. But and that will happen. So I'm not one of those boosters who says, oh, it's just gonna augment every job and everyone would be better off. Like, no, some some jobs that that exist now and occupy a whole person won't exist in one or five years, but most of those jobs are jobs that probably could have already been automated by prior waves. So I think if you're investing in a company and because because of where private equity comes into the life cycle of companies, like legacy is there by definition, uh, not just in ways of working, but in the software products and at some of the events that um you and I have been to and met people at. Private equity people have a hard time, understandably, understanding how technical teams, uh, which they often call RD teams, which is very interesting, do their work and what value they get out from changes in software given what they put in. And AI does sort of threaten perhaps to add a whole uh annoying smoke screen on top of that and make it even tougher to understand, okay, I have just, you know, my bid was the winning bid to take over this payroll company, or you know, I tend to give B2B software examples because they that dominates our customer base, and many of those customers go through private equity uh life cycles. Yeah, there is a there is a risk that the smokescreen from AI further covers like the detail of like how does this thing that I just bought actually work once you get inside the machine, given that from the outside, before you've invested. So I think uh understanding the technology at a very basic level is required. Otherwise, not only will companies that you ordinarily would have bought, right, you might get confused about how they're generating value with AI. But also, if you are then looking to invest in companies where AI is one of their things, you know, you'll sort of be doubly on the back foot. And it doesn't mean that you need to become a programmer and an expert in machine learning and build your own AI models. The good thing about how general purpose the technology is, you just need to be good using it in your own job because the same questions you'll get used to asking are the same questions you would ask anyone who says, I want to sell my company to you, and we use AI to do XYZ. If you've been using AI to do any task within your job, the same questions that about that helped you improve that task are the same questions you would ask to people trying to sell you an AI flavor of something. So yeah, I think the most urgent would be if you're investing in very techie companies where AI is or will be part of their value proposition. Like that is urgent, because you might buy the wrong thing, you might buy something with no defendable moat. More of a problem in startup land, but that problem is going to come to other other kinds of companies as well.

SPEAKER_01

Yeah, there's the other sort of dynamic is well the the you've got a numerous portfolio companies in different sectors, slightly different models. Obviously, there's something that brings them together, you know, why they're in one fund, I suppose. You've also got the fact that that they are at different times in their cycle. So they've just been acquired or they're in the m middle part of the cycle, or they're looking now to to to to exit. And so that will change the way that you know that a uh a P firm will want to how active they'll be, I suppose. Um just does that sort of present an opportunity, do you think? Particularly for I suppose the portfolio company, perhaps an average private company wouldn't wouldn't have. Yeah, can you see it at an upside?

SPEAKER_00

Absolutely. I think um, and as you say, a portfolio company that tends to invest in companies in a similar area of a of a market, uh those companies may be at different um stages. But if they're all payroll or business process kind of companies, or they're all travel companies, or whatever it might be, the thing that ties them all together beyond where they are in the market is the data that they do collect, right? So I think that is a totally unexplored frontier. Um, and AI doesn't do anything totally special there, it doesn't help us avoid the hard work about thinking what data we want to join together and how we want to productize it. What it does do is lower the cost of a lot of that work in a way that we didn't have before. So me having gone through you know productizing and attempting to productize data in the big data era, just doing the work was very slow and expensive. But I think if you know if if you're invested in portfolio companies in hospitality or healthcare or any industry across your portfolio, your companies are are generating data assets that for their limited view of the particular part of that value chain that they serve, maybe they have no other use for that data. But if you could uh unlock you know the commercial value and even having your portfolio companies sell their data to each other, then you might find totally new product propositions, spin-offs, new value. You might launch that thing towards the end of the cycle that helps you have that exit that you want. Uh, and if you do that just in a commercially sensible way, then you're not becoming a sort of uh blocker by becoming a central data store. You're setting up commercial relationships between companies that could increase the value of all the companies. And I think the the what I would really love to see AI do is to help us detect uh you know patterns and useful ways of using data that's generated across those portfolio companies. Because the data that's generated in a given portfolio does represent some of the domain knowledge that is in the companies and maybe in the investors' heads about what works. Um but it's it's records about what products work, it's records about what customers do, it's preferences. Like there's some really great ideas in that data that only someone from the outside would see. But if you're a portfolio company, you have the benefit of having insider outsiders who are not fully outsiders who could sidle up to their to their buddies in the portfolio and say, oh gee, yeah, we we have the same owner and we both work in healthcare, and it turns out the data I've always wanted to launch this product is the data that you have. Can we can we reach uh an agreement? Um so that's a lot of human work, that's a lot of detailed data investigation. I'm excited to see how AI just lowers the cost of that hard work and and allows more companies to do that and think about that in a sensible way.

SPEAKER_01

How um how have you seen how have you sit seen a business? You know, it could be a hire, couldn't it? It could it could be a hire of chief data officer or someone that's going to look over it. It it could be a working group. I'm just trying to think of practically what could private equity firm, how could they start thinking about trying to surface some of that data and not just think isolated portfolio companies, but actually start to think. I mean it feels like a bit of a leap, I have to say, with the conversations I've had at the moment, but that's a really interesting thought where they could start seeing how there could be connections between the portfolio companies and there could be a lot of value in the fact that within the group um than yeah, just looking in isolated cases.

SPEAKER_00

Yeah, the way I would structure an approach to that is you know, I'm reminded of a a recent dinner I went to um with a lot of yeah, investors and companies looking for investment, and a a a guy with some kind of platform in the healthcare and aged care uh industry stood up and said, you know, I'm sitting on a ton of data and I really want to do something with it. Will AI let me do something with it? So you need the commercial person who wants to make money in that conversation. I think many such exercises. So I've the kind of exercise that you're hinting at, I've seen it done within companies, um including by data teams, and I've been part of those data teams. But when it's a technological exercise, when it's a cataloguing exercise, A, it's boring, b, no one corporates, they think you're the police, right? And C it's not being driven by people with a commercial incentive to then invest it. Um, where I've seen it done particularly well is the parent company of the UK subsidiary I used to work for was uh privately owned and it was actually a conglomeration, right? Within those companies and with their external companies, with their uh external companies including their competition, they bought and sold, I can't say the number, but the amount of data that they bought and sold every year was massive, but they bought and sold approximately the same amount in money terms, but that made possible integrations and new products that that weren't possible without it. So if I were to put people in the room, let's say there's three portfolio companies, two might be challenging, we might need three to find something. We need one person from each portfolio company who talks to the customers of that company daily about their domain challenges. So in the sales process, in the support process, what are those people actually struggling with? What are their top three challenges, regardless of what you sell them, right? Which is not a natural thing to always ask. So once you have those people, uh, and then you need, I call this person, yeah, the person who knows where the bodies are buried, right? Uh, someone with enough knowledge about the data estate of each of those um uh companies, and you need some kind of facilitator, right? So that could be an external person, someone like me, or you could hire an appropriate internal person to your portfolio to to do that, but you don't want to hire, in my opinion, a 10 or 100 person data team to just start building prototypes because 99% of those will uh will just uh fail, not even technologically, just if the if the businesses that are supposed to sell that thing. Because what you're trying to do is have your portfolio companies invest less in their core business to invest in something that you want that's bigger than all of them, and unless the commercial people are there from the very beginning and they connect all of that work and the cost of that work to a new thing that will solve a new challenge that they don't currently solve for their customer, but but would increase the value of what they do to their customer. So, yeah, you need the customer in the room through the the salesperson, but it needs to be tempered by the people who know what's feasible with the data that you have, because otherwise they'll just and I've been part of that process too. Salespeople promise whatever data product. Well, we did actually collect that data, so that's gonna be real fun uh to put that. So you don't want that to happen each. But people would probably be surprised how far you can get with a very domain-driven, customer-centric conversation for unlocking this, uh unlocking this data ideas.

SPEAKER_01

Yeah, yeah. Yeah, yeah, it's really interesting. Let's get into perhaps a more practical example of an application that that could help our listeners just try and connect to, and that's what they're asking for. They want to try and connect to to case studies, use cases that that are sort of proven and and have gone through a cycle and have had an impact. Is there one that comes to mind that would be good to sort of talk through, ideally from a non-technical, yeah, not too technical perspective?

SPEAKER_00

So the one that comes up in every training I do, every customer conversation, every industry, every person's job, everyone who works with a computer at some point in their day-to-day work has to take a body of information, right, and and slot it into some boxes. Right? So whether this is here's a whole bunch of text data and I now need to fill out this form, or whether it's here's a bunch of incoming Um investment briefs or chatter on the internet about companies that might may or may not be for sale. That's the input data. And then the other input in this pattern in private equity would be well, what are we actually looking to do? What do we want to invest in? What are our investment criteria? The exact same pattern comes up whether it's software engineers trying to update code from this standard to that standard, uh, infrastructure engineers having to update terraform templates from this to that. Everyone's got some part of their job where they've got a large amount of semi-structured sort of data, and they need to kind of confront that with a a structured standard or or target that they that they want. And so uh you know, depending on which particular kind of PE firm, they might be the kind that um looks at lots of public or less public information and and and has lots of things in the pipeline and they need to quickly whittle them down. And you've but you've already got a statement of an investment criteria, it's probably even on your website. But you even you have a version of that internally. And if an individual just learns how to prompt a large language model effectively, they can very quickly learn how to structured output in the in the the format, the tone, the extensiveness uh that they want based on very unstructured input. And this is not at all revolutionary, humans can do this, and the efficiency gain may not even be wild, right? Because I bet a sufficiently expert uh investor can scan a hundred slides of some document and they've got some investment criteria in their head, and just by scanning it, you know, that the human brain with domain expertise model inside it is incredibly fast at this. But the reason I would push people towards something like that is it's a very good way to understand where AI's limitations and benefits are, right? So it's not gonna double triple the speed at which you can act on an incoming investment brief, but it's a very, very good way to very rapidly increase your understanding of what AI can do. So I I'm over the past few years I've shied away a bit more for recommending upfront as a practical thing, trying to find that one magic super efficiency gaining thing, but to actually um uh yeah, ask people that the first things they do with AI are things they can do themselves. Um, and they you know maybe they get a 10-20% speed up, maybe once they get really good at this, they start spotting opportunities in companies they might have dismissed if they were scanning them. Um but they learn what AI can actually do and what ingredients you need to make it successful, and then the good thing is, if they then start to believe that AI can help them, then they can think about scaling that up, right? And think about okay, well actually, can we scan the internet for this if we transcribe all of our conversations, if we use our CRM and our history, our unique data on the whole, can we, I don't know, uh, go back in time and replay some of the investments we did or didn't make and see what we could have missed? And then you get to the real value, the sort of disruptive value of AI, but you're only gonna trust it to do things like that and confront it with that kind of sensitivity of data if you if you believe that it can spot a cat in a picture of dogs, if you if you don't do the equivalent thing in your domain, you are expecting too much of it uh in one go.

SPEAKER_01

How um with a lot of experiments, proof of concepts, how have you got some practical advice of how you should be measuring success? Um you've talked about you know efficiency and potentially there's commercial upside or there's improvement in um it's it's picked up potentially mistakes or it's elevated your performance. How how in reality you measure that?

SPEAKER_00

Yeah. Um it's easy and it's not easy. Going back to what I said about you know, this um this technology often just forces us to do hard work that we may have avoided. So there'll be all kinds of cases where trying to use AI to do something reveals some inefficiency and actually creates more inefficiency in in the short run. Then so so efficiency is something that I think is best measured by the individual in their own individual workflow, right? So if if you want efficiency gains from AI, you should enable all your people to work with it and to find their own ways of becoming more efficient, right? Because people will do the same tasks slightly differently, they all have slightly different roles. I don't think P firms have you know large customer help desks sitting there doing nothing. They don't have a lot of the obvious places that people would go with AI and and which, truth to be told, other waves of technology have already sort of cleaned out a huge amount of that in efficiency. So I think it's it's too early to think about massive efficiency gains from any flavour of AI in the way that we work, unless every individual in the team that you then want to think how the team becomes more efficient, if every individual has found one efficiency improvement for them in one task. Them simply sharing those will mean that the other people in the team who do similar things can immediately adopt that same thing with almost no adoption barrier, and then they'll come to this thing of like, which is you know partly the whole point. Oh yeah, we made you know these 1990s processes 50% more efficient, and then we realized why do we run it like that at all? So the big efficiency jump will only come from actually reinventing a process, but you need to know the process that you run right now, you need to have tried and failed and maybe succeeded to make some small improvements to that, and that will force you to do that hard work of understanding that process, what data do we have, what kind of AI do we need to help that, and then more even more importantly, even if we had the perfect AI thing at that step of the process, could we integrate that into the rest of our business? Right, and that's pretty quickly gonna become a barrier. Some people are gonna get really amazing at prompting and uh becoming super individually efficient, and they that value won't be realized because uh they'll be pushing against a brick wall. So efficiency is a tough one and it starts with the individual. Uh I think that the another way to measure it is how I can use AI to stop you making dumb mistakes. Not everyone likes to admit making mistakes, whether they're dumb mistakes or justifiable mistakes, but again, you're lowering the bar lowering the bar for what you want AI to do, right? If it's if you find a way to make it double-check everything you do, and if you instruct it in a way to be particularly diligent about checking this or that, then it can be a very valuable, you know, backstop and last line of defense. Again, by doing that, you would then learn what its capabilities are to make it be more proactive, right? And to recommend things that you do rather than recommend things not to do. So this is where we are right now. A lot of the efficiency drivers that people, sorry, a lot of the value drivers people look for. If we want to be successful in the long run, we're gonna have to accept a limited version of that, often a negative, like a flipped version of that, to actually learn how to use this technology before we go go and try and do a big big thing. But if we do that work, we will find new things that we absolutely could not have done, right?

SPEAKER_01

That become possible because of Yeah, so you know, it's almost one job at a time.

SPEAKER_00

Absolutely.

SPEAKER_01

But, you know, I do wonder though, make you know, that will obviously then help you better understand. But rather than just moving on, well, I've done that one, I'm gonna move on to the next one, is that you know how how how important is it that we start to pr perhaps reimagine the way that we work? And not I don't want to be too general here, but it you know, private equity as you know, quite uh you know, there's there's it's quite defined in terms of the process, I suppose. Yeah. Is that is this an industry ripe for I mean this is probably this could be a half-hour podcast on its own. But you know, should they be thinking right now that we probably should have one eye on five years and and how we perhaps look to change the model?

SPEAKER_00

Yeah, I think so. That the challenge is that's a hard prediction problem. The only way to hit the five-year target is to start now in some direction, and the simplest way to start is, as you said, one job at a time. Uh, I can draw an analogy from what you know software went through. So when you know software sent people to the moon and stuff like that, software developers and engineers were generalists, right? They understood what was needed from their software, they wrote the software, they tested the software, they shipped the software, they maintained the software. Then we wanted a lot more software, and when we want a lot more of something, we like to specialise it and say, okay, no, you talk to the customer and then tell these people what code to write, and then you tell these people what tests to run. One of the uh negative effects of this is that the number of people creating stuff at the top of the software funnel exploded because it's very easy to think of things you want to do with software. Um, but the ratio of testers to developers plummeted from one to one because they were the same person to, you know, in most companies that we work with, they're really, really far behind on testing, even if they've if they're doing interesting things in development. They have sort of five to ten years behind often best practice. Uh, we're recording this on the day of the crowds clouds crowd strike outage, right? So potentially one developer put one line of code in a thing that has brought down half the world. That's a little bit of a problem. Now, uh one of the ways that software has pushed back to increase quality and to reinvent processes is to uh do what they call what we call T-shaped skills. Right? So if you're a developer, what do you need to learn about testing to be a better developer and to make sure your testing colleagues can become better developers and most importantly do the hardest testing work? So I think if any industry, and I think P would uh also qualify, if there are like very rigidly defined boxes of humans that do things in a deal flow, the natural way to think about reinventing that is what are the T skills that we do. So if someone only does this part of it, what is the value add that they can do before and after their part in the chain? And then how do you use AI to make to lower the lower the barrier to them doing those things? And that's when you start to get flexibility in your process once you have turned it from an industrial factory, like literal Adam Smith, this person does this to the metal, this person sharpens it. This but you know, humans love to break things down into these incredibly detailed boxes because we think things can scale, uh, and they can if you're making pins in a factory in Manchester, but that's not what private equity does, and that's also not what software development is. Um, so uh you know I've we've had some conversations with some legal firms going through similar things. Uh, any human-intensive, human knowledge-intensive, overly boxified process will be right for change. The the way I would recommend is you know, for each person and each each box, how do you blur the boundary? Now, for many organizations that's terrifying because they like boundaries, but if you want to change how you run your organization, that's uh that's what's required.

SPEAKER_01

Yeah, well, exciting exciting times for those forward-thinking firms.

SPEAKER_00

Absolutely, yeah. And it will be it'll be smaller and medium-sized ones that do that first because they realise, oh, we've always wanted to compete with that big name, true of any industry. Oh, we've always had this constraint. Oh but if we changed our process, we'd get ahead of them on this part of the process, which would then mean it wouldn't matter so much that we were smaller here. And that's that's what competition does. It rewards creativity and AI can help us be more creative if we want to be.

SPEAKER_01

Yeah, yep. Sean look, um really enjoyed the conversation. Thank you very, very much for your time.

SPEAKER_00

Thanks for having me. Great conversation, thank you.