Ties Boukema: Why Relationship Data Beats AI Sourcing Models

Speaker A: Hello and welcome back to the AI Pathfinder for Private Equity podcast. AR Pathfinder helps private equity firms make sense of AI and make it work. It's an expert network, uh, 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 and New York, please check out the show notes for details. This is the fourth episode in our AI Operator series. Today I want to broaden the lens slightly and bring a venture capital perspective into the mix. Because whilst the deal profile is different, the challenge of building AI capability inside an investment firm and across a portfolio is going to be somewhat similar. My guest is Th Bokema, head of data, uh, tech and AI at Dawn Capital. Dawn is one of Europe's leading B2B software investors. What makes dawn particularly interesting for a conversation like this is that they've built proprietary internal AI platforms that have been publicly recognized, including a Forbes feature on how dawn is using AI to supercharge relationships. So this is a firm that isn't just talking about AI, AI native operations is genuinely building them. Tease Himself brings an intriguing background studying law with time at Deutsche Bank, Google co founder and data and AI training firm Cognitas, and now three years plus at uh, Dawn Capital, where he's built out the data, tech and AI function from the ground up. Tease. Welcome to the podcast.

Speaker B: Thank you, Steve. Great to be here.

Speaker A: Fantastic to have you here. Looking forward to getting into the. Into the conversation. Look, first of all, want to learn a little bit more of your backstory and your route into venture capital. You've studied law you five and a half years at Google, you built a data and AI training firm and then you joined dawn in early 2023. Could you tell me a little bit more about that, that journey and how that led you towards VC in Dawn to kick us off?

Speaker B: Yes, minor, um, correction. I did the law degree and a stats degree, which M makes it slightly more logical. I think that I ended up, um, on the data side, better question is why I did a law degree to begin with. And the reason is my dad is a lawyer. His dad was a lawyer. His dad was a lawyer. So When I was 17, I had a deep talk with my dad and he was like, you know, you can sort of study whatever you want. You can do intellectual property law, you could do M and A, you could do real est, but as long as it was law, kind of. I broke his heart and did not become a lawyer. I did two internships in investment banking where I liked many aspects, but from a tech perspective and sort of process innovation, it was a lot less exciting to me than some of these big technology firms, particularly Google because Google also did lots of stuff that was not purely commercial. And the main thing that drew me was not the driverless cars but was uh, that they had teams trying to apply AI and machine learning on medical scans. So eye scans, breast scans, breast cancer scans, skin scans. I've had a brain condition since birth called hydrocephalus which means basically have too much water around your brain. And my condition was very severe as a kid and I've been kept overnight in a Hostel More than 200 plus nights in my life. And I was basically a very, very, very sick kid up to age 20. I'm 65 or like 196 meters tall. I was 165 pounds or 67 kilos when I was 20 which is not a great look. It's hard to go shopping with those dimensions. And that was fixed at age 20 because a doctor diagnosed me correctly and then I had a surgery and as a result it seemed really exciting to get to work at a, at a place that also applied some of their brightest minds on this. So I spent five and a half years at Google and various data machine learning roles, always trying to become technical enough, especially on the data side to be able to get into a team at Google Health. When I then was lucky enough to do a rotation there, it wasn't as exciting or fulfilling as I would have hoped because I think there's many things nice about the Google brand, but the uh, Google's health division I think is a little bit hamstrung. Exactly. Because they're so associated with Google Ads and the headline, you know, Google can see your, you know, prostate cancer screening images is so terrifying for both sides. Right. For both the hostels that would give this data and for Google widely that they move slower than, than some startups would. I then had my first quarter life crisis because I thought this been my, my calling in life and this is where I was going to work until Oprah Winfrey would call me up and ask me to do a book and do the show. And then uh, a mentor of mine said hey, there's this team in, in our finance division that is an engineering team. There was this whole acquisition that we did where none of the reporting and the revenue forecasting systems are led, are created from an engineering perspective. They're all basically built by accountants and very Excel heavy as a result, it just doesn't scale because the scale of Google is just nuts. I mean the team that I ended up leading the revenue forecasting and reporting and making the slides for the, and et cetera in EMEA 4 was about $30 billion of revenue, which is, it's not going to scale with uh, Excel. And whilst sort of setting up the plumbing correctly for that function and making that more in line with how regular Google Ads or hardware, uh, or the other parts of Google basically ran their finance team. I was at a dinner with a bunch of private equity and VC friends and then I realized that, you know, a couple of them had to leave dinner early because they had to triple check like three portfolio companies that they sat on the board of as an observer for quarterly reporting. And my jaw just hit the floor. I'm like, these are massive PE firms that could easily build this in house and should build this in house or they're cutting edge VC firms that have like invested in some of the cutting edge data companies that do this and the software companies that do this. So they clearly both believe in the power of big data and software so otherwise they wouldn't put their dollars there. But then internally they just haven't adopted it in a way that, you know, the average sort of series B company has approached this problem. I think I was even more Dutch and outspoken then a pretty blunt Dutch sort of brand where I was like, well, only reason you people work so hard is because you guys aren't smart enough to use enough technology. And I think there's some nuance I've learned over the years, but there's some truth to still, which is that technology can just scale infinitely better than humans on many, many, many applications and something as dry as portfolio reporting and monitoring. And that is a great example where if you think about it, tech first, you can make a massive dent in it. I, um, interviewed with a couple of firms, had a really, really good connection with dawn culturally because dawn was the only firm really where when I mentioned, hey, I've set up a similar system at Google that would be very applicable to your quarterly reporting. I know I'm on probation for the first three months and I know I'm not going to pass probation if I don't deliver and I want to go do this for you a US firms were much more risk averse where they were like, ooh, but you know, the other prestigious Sandhill funds haven't done it. And like, if they haven't done it then it must be really difficult and it must not work. And I'm like, but if it works for $30 billion of revenue, I think it'll handle your 80 portfolio companies. And they're like, yeah, but we have these really complicated portfolio companies and it's just, it was infuriating.

Speaker A: Yeah, a lot to unpack and we're not going to unpack everything. But what I will say to the listeners is that you sent me a podcast before we arranged this and it's definitely worth a list because you go into a little bit more detail on your condition and the process that you went through and how you sort of got into Google. And I'll put that in the show notes so people can sort of listen to that if they wish to go in a bit more detail. And it's definitely worth listen to. Fascinating. But what I get out of this is you want pace, you want speed, and you're frustrated if that, if it doesn't translate quick enough into results. But the tech is there, you have the confidence and the technology to enable something to happen. And so, you know, if we go then into your role, uh, at dawn as head of data tech and AI, on the surface it looks like quite a broad remit and certainly from the other AI operators that I've spoken to and I just wonder if that shape was deliberate from the get go or has that just uh, evolved as you and they have sort of grown with you.

Speaker B: The initial scope of the role was data, particularly on the data and setting up the architecture because there was no data warehouse and there was very little in terms of tech foundation. Like there was a CRM and uh, thank, thank God they were on Google workspaces instead of pure Microsoft. But there wasn't that much. The big moment when I think the job expanded is when it stopped being how do we get portfolio reporting and how can we get like IR questions answered much faster to enable an effective fundraise, which is probably the first maybe four months of the job into my big moment, I always thought I would actually end up spending all my time on building sourcing models like what you would do at a hedge fund. And I think VC roles are different than PE funds. We're typically on the VC side because we have minority stakes and there aren't just that many dollars flowing out. And a lot of the tech companies you invest in should have their tech in a good place. Right. Like it would be very worrying if I have a chat with an hour with one of our founders and they go, oh my God, I've never thought about this. I'm like, you should have really thought, like, this would be a huge red flag, right? It's part of the dd. Whereas on the PE side, that isn't as worrying necessarily. So as a result, VC roles tend to be almost entirely internally focused, and a lot of these PE roles tend to be very heavily portfolio focused. So the big thing I thought I would be doing once I got sort of some of the core plumbing and finance data in shape, is data driven sourcing. And it took me maybe six months or five months to realize that that was not the most exciting thing to focus on for us because we're a series A and B investor. There was a deal in France, believe it was Penny Lane, which ended up doing really well. And we've been trying to get into that deal. And the. Basically, we struggle to get great contact with the founder and to really have a proper meeting with our GPS and where everything is lined up. And we can properly try to convince the founder that they should pick us over other firms. And by the time we finally have the meeting with the founder in Paris, the founder goes, thank you so much, but I took money from one of the big US firms, one of the firms where actually that investment partner sits on a board with one of our founding gps. So our founding GP begrudgingly congratulates this person on winning the deal over us. And three people message him afterwards from that board and say, well, Norman, we've been on the board for many years with you. We really like you a lot. And me and Anna were both angels in this. We've known that CEO for a long time. Why don't you just ask for an intro? So he stormed off to the deal team, and this is the first time I think I've ever heard him raise his voice because he. He's a very balanced person. And I'm, um, paraphrasing, but you could imagine the, the language he used where he's kind of like, why the. You know, why did you ask me for an intro, guys? I mean, we had a really good path to play, but I didn't even know that sort of, you know, the man was open. Like, why. Why didn't we. Why didn't you ask me for an intro? And they're like, well, how am I supposed to know that you are. You're not part of the deal team. You sit on a board with someone who is an angel investor in this company and someone else who used to be the manager of one of the. Of the cto, I believe, of the company. And I'm just Thinking if you just have all that information as you can totally know this because we can actually mine and buy data and stitch it all together to understand who's co invested with who, who's worked with who in what capacity. And was it a small company or a big company? Did they both do similar roles? Was it in the same location? You know, what companies have we invested in, where do we sit on the boards, who sits on which charity board, who are our LPs, you know, our LPs, joint LPs in some GP that is actually earlier stage, which GPS or angel investors over the last couple of years and especially recently actually have been doing deals that are very exciting are rising stars at certain funds. Because I think a lot of the tier 1 VC, tier 2 VC, tier 3 VC is very circular. There's lots of funds that haven't actually done fantastic returns or deals in the last couple of years that people still deem to be tier one investors. Whereas a tier one footballer is the best at their position. And if you used to be really good three years ago and now you deserve to be on the bench, then you're not a tier one footballer, right? But in VC you can sort of limp around the field or have the perception that you belong on the field, that you can pick for the national squad for much longer than I think is valid. So I had this idea that if we could leverage our network incredibly well when we wanted to get in touch with a founder afterwards also we built the capability to get in touch with an LP or with an operator to do background checks or to do reference checks or to vet someone or to an angel investor, basically any, any sort of person or company. How can we mobilize the entire dawn network including WhatsApp relationships, calendar meetings, work experience and not just direct, but even like say if I am good friends on uh, WhatsApp with someone who has done an angel ticket in a certain deal. This doesn't show it in ecm, but it's actually extremely high signal and I think it's a harder problem to solve because even if you say we can just calendar data without doing anything with mobile phone, Cs tend to be very sociable. So if you look at the relationship arc on a two by two, then it starts with having a bunch of zoom or calendar meetings and then at some point you become close enough that you just give each other your WhatsApp and then at some point you stop inviting each other with your work email because it's weird. Like if we've known each other for years and years and we're whatsapping all the time. It feels a bit odd if we're going to go for a run on Saturday to send uh, an email in your business. Email. Right. So I think there's a lot of nuance in terms of how do you understand what is the real accessible network for not just a person but for a firm and their and who they would know. Well but this has been by far the thing that's actually helped us make more money because both in terms of understanding who are the people that need to be real fans and advocates of dawn so that we get access to deal flow better and we get tip offs. Because the exciting thing about private markets is you can get information about a private markets company about for example, hey, this company just signed some proof of concepts and these proof of concepts are doing incredibly well. So these are going to become multimillion dollar contracts in private markets. That just means you're a good investor and you're well informed in public equities, you would hopefully go to prison. Right. So I think private markets is really exciting in that sense where if you do that well, you get access to deals, you win deals. And now whenever we get excited about a company being able to punch in give me 10 paths to Steve Budd and not just have it surface. Hey T has had one calendar or two calendar meetings with Steve but actually there's people that we know that would have, you know, maybe invested or worked together or there's media coverage link taking you and another firm together like this. It's just always, always valuable to be able to serve for the totality of the effective network you have as a firm to a given entity.

Speaker A: Thank you for going through and explaining that. How have the team, the humans adapted to this and ah, how does that look in terms of the way dawn operates? You know, has it changed beyond recognition? Be good to just understand what that looks like a little bit more.

Speaker B: So there's the old way of us trying to get in touch with a founder would be to put the LinkedIn URL of the founder in Slack and then can people Please check their LinkedIn to see if you have mutuals that make sense. Now there's many problems with this. First of all, not everyone is going to check on LinkedIn and even if you would on LinkedIn you miss the calendar data so you have an incomplete piece. Most importantly, second order connections in LinkedIn are so noisy that they're basically useless. Right. Like for a lot of people that are sort of swimming in the Same pool. And if you've been even in venture for three and a half years, for any founder, there's always a whole bunch of second order connections on LinkedIn. The hard part is, well, which of those is going to be good enough that I can actually WhatsApp Oliver or Anna or whoever it may be and say, hey, you guys actually have a logical connection or you guys have actually have a pretty high probability of knowing each other well enough that I can ask for that connection and that even I can ask Andrea, uh, be like, hey Andrea, uh, I see that you did a, the same, you know, student club for two years in a leadership position at Oxford as the chief of staff at that company. Now not just the CEO, but also everything around the CEO or getting to board members or getting to advisors of the company, or getting to angel investment. Like there's actually quite a lot more surface area to begin with than just the founder that is very, very useful and very useful to influence. And even if you would just want to get to any of those people being able to mobilize the whole thing. So in terms of how we try to get in front of founders and even once we have met the founder, it's still incredibly useful to understand all the different ways how can we get either information or how can we get them to pick us over Accel or Sequoia because you understand all the surface area of that entity. So that piece is completely different with Rolodex. We're also much more, I think intentional that everyone now has a clear wish list. We call this Rolodex coverage where we're able to say, hey, there's someone at this fund that is either a rising star or a fantastic investor and if they would be a fan of dawn instead of neutral about us, that would benefit us. So one example of that is um, Begin Capital is a Dutch London based fund. But now love Begin. So without saying anything mean, I don't think they are most people's idea of a tier one firm. If you ask, hey, give me the top 10, 15 early stage investors. I don't think that many investors will actually say hey, I've heard about Begin because their, their brand isn't as good as their investing acumen. And there was someone there called Joelle and Joel had made general partner in very rapid time. So clearly because it's also sometimes hard if you look at crunch based data, et cetera, to figure out who, who at the fund has done a certain deal. But if you know that the fund has done great deals, there's a bunch of deals that don't have clear partner attributed to them. And there's this one guy called Joel who's made GPS in not much time. You know, odds are pretty good that Joelle is someone that's been instrumental for the success of the firm in recent months. Right. Another example of that is Oliver Cakeset um, concept. So by using this data and by using this graph, we could say, well, it would be really useful if one of us would be good friends with Joelle. And you can't force this. Right. With some people you have a click, with some people you don't. It does give us a very intentional moment once a quarter saying, well, actually if we're going to do events, we should start the invite guest list by inviting some of these people where it would be lovely if they liked us more. And that each person at dawn, especially each investor on, should just have certain people that they uh, are responsible for trying to increase the relationship with in a data driven way. And uh, Joel is actually the person that ended up then giving us the tip off that runware was doing outrageously well and that they just hit this massive revenue inflection point which allowed us to preempt the deal. And I think given it's in such a hot space, it's AI inference for video and it's now done, I think 6 to 100 million in revenue, something like this in like a year and a bit. I don't think it would have been a certainty that we got to do that deal over say Sequoia if we hadn't had the relationship and the tip off and got the timing right because we were able to put an exploding term sheet on the table that uh, we wouldn't have felt the confidence for otherwise.

Speaker A: Yeah. So you perhaps become more purposeful in terms of the way that you're operating and, and certainly proactive. I'm just thinking in terms of more practical steps that you took. And I suppose this is just the environment you're in with your portfolio as well. You've been building rather than buying. But was that just the way it's been? Or is, is there been an opportunity to fast track something by looking what else is out there? But I'm, um, essentially going to answer my own. There just wasn't access to something similar. Hence why, um, this was so unique for Dawn.

Speaker B: Yeah. So for, for Rolodex specifically. Um, look, I think it used to be that people were aware that building great software, software that actually sparks joy and is useful and gets used a ton, is Very hard and very expensive. And then now with Claude code, I think the narrative has overcorrected a bit where any 23 year old with a cloud code subscription can build Salesforce or Google in a, in a weekend. And I think there's still a lot of, there's a lot of middle ground there. In general, buying software is virtually always cheaper still I think than building your own and maintaining your own. So there's been a couple of systems that we've either just fully bought or that we ripped, ripped out and replaced because we said, hey, actually we can get a very big improvement in the outcome or of the foundation if we just replace our finance system with a better finance system and we just squeeze much more juice out of the new thing we bought rather than going into full build mode. That said, for Rolex specifically, I still to this day really don't see anything comparable in the market. And it kind of makes sense because private markets is pretty niche and I think the problem is hard and it was a much harder engineering problem than I thought. It also quite frankly took almost a year before we really started seeing decent results and then maybe another six months before it got used a lot and also really got us into a whole bunch of deals. It took longer for sure. Uh, especially ex Google engineers tend to be pretty arrogant. They're like, how hard can it be? And uh, pretty hard is the answer. So where you can buy, I think you should buy. But for something that it was so obvious that it made every core part of our business better, right? It now helps us get warm interest to LPs than we wouldn't otherwise. It helps us be much more strategic about when we plan travel. When someone has a board meeting in New York, who should they meet? Which founders should they meet, which investors should they meet, Which LP should they meet? Rather than making this an afterthought, I mean sending a GP to New York, beyond the cost of the hotel and the flights, just based on their value for the org is an incredibly valuable three days. So it seems absolute lunacy to me that at most funds they sort of scratch themselves behind the ear three hours or like a couple of days before boarding and they just sort of think, oh, who do I know in New York? Like surely there should be a tech starting point that suggests, hey, there's five people or 10 people or 20 people, whatever it is, for these reasons that are exciting for you to re engage or engage for the first time. And this is how you would set up a meeting efficiently. And this also now that system works well enough that also especially route travel, a lot of this work gets done by the EAs, which M makes it even easier for the partners. Right? It's just like they, they have a board travel or board trip, they say, hey, actually around the board, you should ask about those deals to that person and you should try and squeeze in a couple of more meetings. And this because it's so easy to do this, do this proactively. A lot of these meetings now also get scheduled much further out, which means the availability of the person's better. Because my calendar three months, I'm like, absolutely, I've got time. My calendar next week, I'm like, oh God, Steve, please.

Speaker A: No, Yeah, I mean I'm building a network, I'm building a community. And one of the things I want to do is to not just leave it down to serendipity that someone meets someone down the corridor at one of my events and has a really valuable conversation because that may or may not happen. And so, you know, how do you do something a little bit more purposeful and connect people in a curated way that, you know, that's something. I'm doing it some scale, but you know, to the scale you're doing, I mean we're talking about decisions that could mean a lot of money. And so it's highly, highly important. The one thing you have sort of found through this, and you talked about your three month probation, is that there was perhaps, uh, a rush to show some form of ROI so you could keep your job. But there is definitely out there at the moment. There's a frustration growing around the lack of quality ROI for all the investments that are being made. It's like there seems to be this sort of dual line at the moment is that uh, while we're spending more, but not much has changed in terms of cutting costs or additional value.

Speaker B: Yeah, margins for sure.

Speaker A: So obviously I'm talking more on the P side here, but what's your experience been? Perhaps wider. Uh, and should we continue to be patient?

Speaker B: I think you should always have a healthy tension between patience and impatience with these things. I do think that the way that operational AI has been adopted in some engineering functions where a lot of the thoughtfulness has gone out of the window and a lot of the deep thinking has gone out of the window. Window because just slap it in Claude code and then if Claude says no, then I guess the answer is no. And I think there's a lot of people that over rely on this. Maybe one example is when I really have to write something thoughtful and original. This might be too personal because I've only been married for a month, but I write poems to my wife once every month or so. And on paper you'd be like, well, writing is something LLMs are really good at. Ah, I could just feed my previous poems in and then in no time I could romance her over and then I. I wouldn't spend so much time doing this. But I still find that if you really want to write something thoughtful and write something well, there's a just this AI style of writing that is easy to pick out and gets very annoying. You see a lot of it on LinkedIn where everything has this like, punchy first line and punchy last line in the paragraph. And it's just like if any human would write like this, someone would hopefully slap them. Actually, it gets very annoying after five minutes. So I think for some stuff, you want to carve out time for actual deep thinking. And I think this applies to writing, which is why I still write these poems by hand. And same thing with a lot of the coding problems. When I really have a hard coding problem, I will actually walk away from the laptop just with a piece of paper and I only get back to the laptop when I really have a plan of action or I have a clear thing that I'm doing and then I force myself to only do that thing and then walk away again back to my piece of paper. I think there is a floppiness of thinking that has happened today because of AI and because I think historically a lot of text you read on the Internet, like on Google, Google used to be quite reliable. And I think now with AI, it looks, you know, like text, therefore it must be true, and it's just not the case. That said, uh, I think if you pointed at the right problem, I think the gains of AI and technology widely are just massive. Right? Like two examples. I mean, the most exciting thing is by far Rolodex in terms of actually helping us make money. Because there's at least three deals that were surfaced. Accessor Won, or all of the above because of Rolodex. There's been a couple of deals where we were able to uncover ways to DD the extended founding team and uncovered some real red flags that meant we did not do that deal, which I think is also very valuable. And then on the operational stuff, quarterly reporting. Now, when I joined Dom, um, it was about. There were 400, 500 comments between the spreadsheet and the slide deck of people going, this number's wrong. No, it's not. Yes, it is, but it doesn't line with the boardback. Yeah, but the boardback actually uses LTM revenue. This uses endless. Right. Last time there were six and actually the numbers were correct. It was just apples to oranges with one boardback. So being able to prepare this kind of stuff and then being able to also analyze our portfolio and understand what is happening to market margins in certain companies, what is happening to revenue growth compared to two years ago, what is happening to our cash position and Runway in a way and being able to interrogate the data and the same thing on the IR side, being able to a just have IR decks that are correct all the time where there's no one needs to be chased about, is this the least commentary? Like that whole workflow is just fixed. And being able to answer DDQ questions drastically faster. We were able to turn around a DDQ question because almost every question asked actually there was a variation that was asked before. You know, it's not just a quad skill but like we built our own thing. It doesn't work that well out of the box. But the end result is that now we can sometimes turn around a DDQ in 24 hours and that's with both the IR VP and one or two of the GPS checking the answers personally and being happy with those. And because our whole spiel now is, hey, we operate better through AI and technology, I think it would really undermine our case if we couldn't then also turn around the DDQ version very quickly. And being able to turn around these DDQs builds a lot of trust with our LPs and helps a lot of these processes move faster. Right. I also find it funny how on the deal side every deal team is like, well, obviously a lack of speed is going to kill deals. But then on the LP side people like, well, it might take two, two and a half weeks to get back to the question. Two and a half weeks, That's a long time.

Speaker A: So ROI is nuanced and in this sort of phase that we're in at the moment, there's a balance to be made to be had. But pe, especially, you know, their portfolio, portfolio, they are wanting to see what they would normally want to see, which is margin ebitda. And so it's something that they're really trying to squeeze out. Now, do you think this is a case that we haven't yet found the playbook yet? Is it? Because we haven't really got the operators that are able to diagnose and understand and prioritize where the portfolio should be Heading, do you think there's an answer to this just yet or do you think it's a continue assessment and we will get into it in a moment in terms of building out the capability that can support portfolios better to try and identify because part of this has been around people as well. And a real gap that the PE is seeing is in mobilization is people. And so I just wonder if there is something that we should be focusing more on that's going to deliver a little bit more of the answers to that ebitda, uh, question.

Speaker B: So I'm very bullish on tech and I think there's very few PEs and PE people that I speak with and I'm going to maybe add a big legal disclaimer here that I currently don't work in PE and I haven't done value creation with PE companies. So again, shooting slightly from the hip here, but I think what is a fact is that PE companies themselves have been pretty awful adopters of technology generally. And I don't think that the senior leadership at these PE firms has typically gotten promoted based on their expertise on technology and machine learning and AI and data and all that kind of stuff. And the opportunity for technology to do many, many, many of these processes just drastically better to me is irrefutable. So I'm very bullish on what's possible if you get the right operators in. I think getting the right operators into a PE style environment is tricky because PE culture and tech culture clashes quite aggressively. PE management style for an investment associate and for an engineer typically doesn't lend itself very well. The whole move very fast with relatively little supervision and potentially break some things lends itself very poorly to PE think, especially on the internal side. A lot of the PE projects have to happen quite quick, like a deal moves quite fast, generally speaking. And a lot of the internal projects move quite fast, generally speaking. When a tech company says, okay, we're going to build a new capability or totally new product. From the our case with Rolodex, it took probably three months, six months to have a thing I could call Rolodex and then almost a full year of hard focused development where I just do internal tech, right? And I have a team and this, this is really the Rolex have been 80% of what we do probably took a year and a half before I was like, yeah, okay, now it's clearly demonstrably an advantage for us, an edge that has gone us into certain deals that is that I can tie to money, not like time savings or whatever, because if you give people time savings they might just browse Instagram a bit more. Just think play. Often you don't instantly see it in the margins but really where I could say we've now built a piece of tech that other people do not have that is only going to get better, that is making us more money. And I think that culturally is very hard to do in a PE environment which is why not that many funds I think have done this very successfully. I think there's some exceptions. I think HG Capital is one of the very best at this. Some of the people that I met from the CBC portfolio value creation team I found very strong and impressive and I've heard good things about. But I think in many ways to do this capability well you can rely on external support but you also really need the DNA to change at senior levels in the organization and ideally also not be a cobbler with broken shoes.

Speaker A: I think that's where the rise of the AI operator is coming and seen it amongst the network that there's been this rise of this role that sits in the P firm um and they seem to appear to have most of them a dual role of internal initiatives as well as supporting the UM portfolio as well. We are coming to time and I just wonder if we could just perhaps look forward a little bit more maybe 12 months, whatever. It's really hard to think too far ahead but it'd be good to know what your sort of focus at dawn is over the next few years months, what that that will be and perhaps then also how your role will adapt as this sort of moves forward. How it's adapted perhaps up to now as well because it could be that we're the investment firm does things differently that this role isn't necessarily needs to be dedicated over time but yeah looking forward and how the role might change would be really good way to close this out.

Speaker B: So I think as a rule of thumb I think it is much easier to do one thing well than to do two quite different things well which is why if you are a think even mid sized fund and Don is large for a VC with 2 billion AUM but would be tiny for a uh, for a PE I think it makes total sense and I think you should have at least one person fully internally thinking about this, wrestling with this, building solutions and driving them home and iterating on them just because I would not have been able to build Rolodex if I also had to do deals if I was a 5050 role or I think it's just Very distracting. It's very hard to get the balance right to do end of year compensation and to do annual reviews. I think in private markets relationships are an incredibly valuable asset and I think the around how do you leverage these relationships? How do you figure out where you want to nurture? How do you figure out who are some of the highest leverage people that your firm is not close with yet? I think is and remains a uh, superpower. And I think there's something around relationships and relationship intelligence and networks that I think is just totally undermined by private markets as a whole. I think there's a lot of work already seeing this at dawn now where you know, we built our own plugin in Claude that has a whole bunch of nested skills and assets and we built their own MCP as well for many services that basically helps you just use Claude for all the daily things you do much, much, much better than it does out of the box. So it is much more precise when I say look at Fuse Energy to pick the right company. If I ask you to look at Steve Budd, that it is much more precise about finding the right Steve Budd and all the associated data that we already have in our data warehouse. Uh, that it can actually make slides that are used, the right font, the right style, the right format, the right build up, the right structure so that actually anyone can see. Talk into Claude for 3 minutes, 5 minutes and it'll have a slight outline that is just very strong or throw an existing deck in and get kind of partner level critique and suggestions on how to make it better. I think is a superpower and being able to do many parts of the diligence substantially faster than we could before. I think there's a huge amount of the operational work around deals that is going to change a massive amount. And then I think in the back office of these firms, whether it's quarterly reporting or LP reporting or legal stuff or answering LP questions, I think there's a way to think about virtually all of these processes, especially the scale of HG where you can really design them. Like how a tech native firm or an AI native firm, honestly closer to a Google than closer to a place that lives and breathes Excel every day, which is very exciting.

Speaker A: Well look, I feel like one piece of advice, you know, certainly if a firm um, is looking to hire an AI operator, it seems to be, is just be mindful that just bringing someone to look across uh, internal, external is going to be really difficult for them to put a focus and they'll assume they'll discuss get pulled into the deal discussion yes yes. So that that focus is is. Is going to be pretty key but you know for some this would be their first hire anything like it so I suppose there is some form of learning curve to be done but yeah, that focus is going to be essential to ultimately get the ROI that you want T It's been fascinating conversation. Really, really enjoyed it. Thank you so much for your time today and it's been a usual and um, valuable counterpoint to the, you know, PE focused episodes that we've had and previously. So thank you so much.

Speaker B: My pleasure. Thanks for having me.

Speaker A: Ste more about Thiece in the work he's doing at dawn and we'll put your details in the show notes. But that's it for episode four of the AI Operator series on this AI pathfinder for private equity. If you're in a similar role, whether in PVC or portfolio company, we'd love you to take part. So give us a shout. But that's it for now. Thank you for listening.