Subscribe
Share
Share
Embed
The "manual era" of capital allocation is in its final chapter. The firms still relying on manual data extraction and analysis aren’t failing overnight, but they are falling behind one week at a time. While most of the industry continues to "white-knuckle" through 200-page documents and legacy databases, and manual Excel extraction, a new breed of Agentic Allocators is quietly rewriting the rules. They aren’t just using AI to summarize emails; they are leveraging AI-augmented workflows that intelligently automate parts of their investment and operational processes that were previously impossible to automate.
Hosted by Victoria Sienczewski, CEO and Founder of AuumAI, The Agentic Allocator is the "behind-closed-doors" look at how the world's most sophisticated Limited Partners (LPs), allocators and General Partners (GPs) are actually deploying AI, and the hard-won lessons from those building the systems.
This isn't a series about high-level theory or technical gibberish. Each conversation features industry leaders, forward-thinking LPs, GPs and experts who are rewriting the rules of capital allocation through agentic AI. Expect real-world case studies, tactical frameworks you can actually use, and moments that challenge outdated norms. You'll come away with a clearer understanding of the critical questions every allocator must ask - about data privacy, team adoption, integration, and governance - before investing in any AI solution. If you're tired of the "black box" and ready to evolve your investment office for what comes next, you're in the right place.
[00:00:01] Victoria: Welcome to the Agentic Allocator. I'm Victoria Sienczewski, CEO and founder of AuumAI. If you're an LP or allocator, you're likely seeing two different worlds right now. One, where your team is still copying and pasting from 200-page LPAs and DDQs, and another where Agentic AI is supposedly doing the manual grunt work and analysis for you. Today's guest is someone bridging that gap, a practitioner and expert who can speak to what actually works, what doesn't, and what it takes to build trust in these systems.
[00:00:36] Victoria: Welcome to the Agentic Allocator. Today, we're joined by Breanna Genecov and Kunal Koppula of Makena Capital Management, a $22 billion outsourced CIO managing multi asset class portfolios for endowments, foundations, and family offices. Breanna, who leads the portfolio solutions and OCIO team, has spent over a decade at Makena. Throughout her tenure, she held various positions in manager research, portfolio strategy, and risk management, ultimately focusing on the design and implementation of multi asset class portfolios. Kunal, who runs Makena’s data engineering team, spent time as a trader and a quantitative developer prior to joining Makena. And now he leads all data related workflows across the organization. Kunal and Bre are at the forefront of a firm wide transformation of Makena's data architecture, moving from a historically manual system to a centralized foundation designed to support scalable analytics and future AI integration. Today's conversation focuses on why data readiness is the critical first step for institutional investors preparing for AI integration.
Welcome, Breanna and Kunal. Thank you for taking the time to speak with me. Really excited to dive into the data project and what it means for Makena in the future. And we'll start off with focusing on the specific moment or catalyst for Makena where you realize that data was the first step to everything you wanted to achieve, including AI integration. So curious, what was the initial catalyst?
[00:02:05] Breanna: It’s interesting. I would say it's hard to pinpoint an exact moment. I think there's always been an awareness that our systems could be improved and kind of taken into the twenty first century or beyond. Makena was founded in 2006, so our first big data project was in 2016, I'd say about ten years in. And we moved into a platform where we could actually do more analysis in the system. I would say adoption of that was challenging, though. Certain people adopted it, while others still did the download into Excel and used models to do analysis. And we're now basically at twenty years, and we're doing our next rev of this. We invest in some of the leading technology companies around the world, and so it's only fitting kudos to our leadership that we're trying not to be left behind when the financial services industry is often laggard in adopting technology. And this has been a push and a building over the last few years to really try to be at the forefront rather than playing catch up over time.
[00:03:08] Victoria: It makes perfect sense. And, Kunal, I know you are leading that data project internally. What does it entail? What are the steps?
[00:03:16] Kunal: Yeah. It's a complicated question. Data touches every team we have at Makena. So everyone is looking at things slightly differently, but, ultimately, in most cases, using the same sources. And so not only do we need to adopt a whole new set of tools, like tools based on SQL, based on Python, but we need to also, like, upgrade the way we think about our operations and the mechanics of everyone's day to day job. We're about a year and change into the project, and we started with scoping out what is necessary, how hands on we want our tooling to be, what kind of skills we need our team members to have, and try to understand, like, how far is this going to go and what departments and operations it's going to touch. And then we started with, okay, let's move all of our data into a singular source. And so I was trying to optimize basically every ingest and egress of data into one place. And finally, we're sort of in the first phase of porting over a lot of our tooling. So things that someone at the firm in Excel would manage on their own and have this Keyman risk to it. We're saying, let's make that into an automated process so that not only will this persist after you're gone, but also that we have all these things running in the background overnight in the cloud so that we can build new things on top of it. I think it's really important, especially at a place like Makena, that there is a very cohesive lineage to everything that we do, that what client operations is building and investment team is building are based on the same thing. We're not spitting out different numbers in different contexts, but we have that unified approach of every tool and every process build on one workflow, one set of data.
[00:05:10] Victoria: Incredible, Kunal. And, obviously, that hasn't been the standard for the industry for a long time in the sense that there's been a huge manual tax that many teams have paid. And curious if you could describe that workflow, that manual tax that the industry has really focused on for a long time? How would you describe it?
[00:05:29] Breanna: I think there are a couple different elements of it. From the very first phases of just getting data, I think our financial operations and accounting team have to pull the data from emails or PDFs or logging into portals and downloading PDFs and then transferring that data into our accounting systems. So that's one piece of manual work just to get the data into its own format. There's work being done to help automate that, but that is still very much a manual task and needs people checking. The second piece is whenever the investment or portfolio management teams want to do analysis, we go into the accounting system, we pull down the data, and then we put it into Excel and try and manipulate it and do analysis to track performance or allocations or risk and all of these other things. And anytime we wanna update that for a new month or a new quarter of data, it's a manual process to pull it down, put it into Excel, and rerun the models. And there's a lot of room for human error in that and just a lot of time spent updating models that could potentially be, the hope is, updated more autonomously if we have the right systems in place.
[00:06:45] Victoria: Makes perfect sense. And curious, when you think about upskilling or retraining people on your team to start to work in a new environment where things are synced, there's no more manual work within the context of their historical jobs, how do you think about that, at Makena?
[00:07:03] Kunal: We want all of our analysts, you know, investment people, client people. Their primary job should be analysis of investments, talking to clients. Their jobs aren't in Excel, moving numbers around, copy and pasting things. And, similarly, in the future paradigm of how we wanna operate, we don't want them to be just running SQL queries all the time and messing with background infrastructure. Like, their job should be providing value for their team, right, not becoming data gurus in a sense. So we wanna give them the skills to interact with the data, so think SQL or Python, without that being their day to day job interacting with it all the time. And so we don't want people outside of data engineering to have to spin their wheels to get the information they want. It should all be retrievable and accessible, and then they can slowly learn to manipulate that data in the ways that they want and then put in automations to make sure that, like, these things get produced at the right time, sent to the right people, that kind of thing. So we're definitely in the midst of upskilling a lot of people around the firm. Moving from Excel to a BI tool isn't always apples to apples. There's a little bit of a skill barrier there, and there's things you can do in a spreadsheet that you can't do in more advanced sort of technical tools. So we're learning how that operates, and I think that a large part of that is enforcing this rigidity that comes with standard data engineering practices is you can do this, but you can't do this. And the reason is we don't want people to be able to change numbers across the firm. Right? You should be only be able to do certain kinds of operations, and that's good for everybody.
[00:08:44] Victoria: Super interesting. And three very important teams at Makena, so investment, operations, and also client. How do you think about making sure that three teams are unified from a data technology and potentially in the future AI approach and strategy rather than creating silos across the organization?
[00:09:04] Breanna: A team that Kunal is leading, like centralized data engineering and strategy function, I think is critical to that because he's the connective tissue that talks to each of the teams and understands their needs and can help us develop the systems in the right way so that when there's overlap, which I think there is a lot of overlap, it's structured so that each team can get the unique things they need, but it also feeds into the needs of the other teams.
[00:09:30] Kunal: I think there is so much overlap that previously people had their own algorithm of how to get to the answers they want. And my seat's very interesting. I get to talk to all the different teams, and then I say, oh, client's doing something very similar. Have you talked to them about this before? And maybe someone did many years ago, but it's been a minute. So in the process of automating a lot of the things that Makena is doing, we also get to notice these places where there's a lot of overlap of work being done on a daily, weekly basis. I think what's really important is that we don't redo the same thing in multiple places. Right? It's a very core data engineering principle, but one source of truth is really important. You might have your own filters afterwards. You might join in with something that the client cares about, but the investment team doesn't. But that initial cut of information should be consistent across the firm. And if it's not, it means that two different teams are operating on different sets of information, which will lead to incongruous decision making.
[00:10:33] Victoria: And what have some of the challenges been, Kunal?
[00:10:36] Kunal: It's been a lot of good challenges. My background is in public markets where there's a lot of structured data, and there's oftentimes a lot of overlap between design thinking and investment thinking. I'm trying to impart as much design software engineering thinking into the processes that have been going on for a long time. I think the people here at Makena have a lot of context of how things should be done and what the right ways to look at the information is while I have the design principles or building software side of things. So we're trying to put those two things together, which means sometimes I say, can we just skip this portion? Do we need this table? And someone will say, no. We need it because this person and this person really care about it, and it helps them inform these decisions. So it's been a lot of back and forth of how do we design something for the long run, and how do we make sure it meshes with the old way of doing things. Because we don't wanna just say, hey. Here's a new tool. The numbers might be different, but it looks good. It's an unusable tool if it doesn't match up with the way people think about things previously, and it doesn't match the numbers. So we have a lot of validation going on. We have a lot of back and forth about what should the process look like in the future. And slowly but surely, we're sort of having everyone learn how to operate with data from a practitioner's perspective rather than a free for all of Excel documents and PDFs and things like that?
[00:12:01] Victoria: It's such an important challenge. I actually think as an industry, connecting the deep domain expertise with the technology side is lacking across the industry. So I'm so happy to hear that you're playing that vital role internally at Makena. Super important. Would love to understand how the data transformation has changed, how teams at Makena are working right now, and how you think it'll change in the future as you continue to evolve and transform the organization?
[00:12:29] Kunal: I think the fun part that we've noticed already is that we have ported over some tools into our new data stack. And immediately as soon as it's done, we've had people go, that number looks funny or that number looks funny, and we're getting this natural discussion going with, like, how we do things and what the right numbers we wanna look at are. And then people are already saying, oh, I really love another downstream table that puts the performance with the NAVs that we're seeing. Right? So people are already saying, like, I wanna extend it this way. I wanna extend it that way, which previously and correct me if I'm wrong, Bre, but there's so much effort until just making the first thing that they never got to the V2 or the V3. Our data stack is set up so that once you have the base case extending with new SQL queries, with new charts and tables is fairly straightforward. So we're already seeing some of that. I would really love the next component, the V3, the V4, And my backlog of tickets is growing exponentially of people saying, oh, I'd love to integrate that data source too. So the snowball is rolling on the hill, and we're seeing people engage in a new way, which has been really great.
[00:13:42] Breanna: Kunal is the most popular person at Makena right now because everybody wants to move their project forward when he has time. And I think one of the other things that I would add is just how the senior and junior teams are working together. Because when something's in Excel, like, the senior teams aren't gonna get into the weeds. Sometimes they do, but, generally, they leave it to the analyst. Whereas now that we have this platform where you can go on and access the data directly, there's more dialogue about, oh, this is great information and analysis, and here's the secondary and tertiary analysis that I want to come out of it. And building that way is a little bit more integrated than it has been in the past.
[00:14:24] Victoria: Super interesting. A little more self-service for the senior team members and then allow them to also ask the junior team members to create new analysis that maybe has never been able to be done before, give the historical kind of data or inconsistencies or lacking from a data perspective. Super interesting. Kunal, we'd love to dive into the phases of the project a little bit more. So phase one, I understand, was purely focused on investment data. Can you tell us a little bit more about that and how you thought about that?
[00:14:53] Kunal: I would maybe add to that and say the first phase that we're still in and will continue for a long time is a lot about structured data. And so investment data, benchmark data, anything that can really be put into a tabular form and operates in a sort of SQL environment. And we have a lot of tooling around here that we're trying to port over, that we're trying to make new tooling just based on structured information from a variety of sources, whether that's our data from our clients, whether that's Bloomberg information, whether that's private market information. A lot of the work we're doing has to do with structured information, but I'm sure you're aware, like, the unstructured component plays a huge role in firms like ours or really any firm that touches the private markets. There's so much detail and work that gets put in by our investment team, whether it's conversations with potential managers, whether it's deal cloud information, whether it's a PDF that gets sent once a month, or maybe it's a PDF that gets sent, like, every three months, but an extra one gets sent in April as well. There's just so much variation to it that it's hard to always put it into a structured way, and that's where a lot of the phase two efforts that have begun are going to live. There's the component of turning unstructured into structure, but there's also a completely different pillar that is make the unstructured available and contextualize with AI. Right? And I know a lot of firms outside of finance are thinking the same thing is I want to be able to access my structured and unstructured with an AI lens on top that can put them together. And we're certainly very excited about that phase. We're very nascent in terms of how we want to approach that. The world of AI is so broad and touches so many things. There's good ways of doing it and bad ways of doing it. We're simply trying to decide what's the right approach, what's the right project to start with and learn some things on so that we can bring the unstructured and structured together the right way?
[00:16:54] Victoria: Well, it sounds like a fantastic approach getting the foundation correct before you dive into automating some of your workflows using AI specifically. A lot of your peers haven't been as thoughtful as you have Kunal and Bre. And so curious, when you think about other peers that, let's say, are jumping into AI right away without having access to that structured and unstructured data, what advice would you give them?
[00:17:20] Breanna: I would say the biggest piece of advice that I would give is don't expect a panacea to begin with from AI. You've gotta actually do the work ahead of time to get the infrastructure, the foundations established so that you can actually leverage the data in a way that AI can use. AI seems to be this savior that's gonna come in and, like, do everything that you haven't been able to do before, but I think there's a lot of work on the front end that goes into making that possible and making it actually robust in a way that's useful to the organization and is scalable as AI improves and as our dataset continuously expands.
[00:18:03] Kunal: I'll just add that a common data science principle is garbage in, garbage out. Right? So if you say, AI, connect to my database, connect to my folder on my drive, like, and then give me an answer to this question. Nothing's been organized, and it's gonna give you an answer. But how it put together that answer doesn't have the context of an analyst. It doesn't have the structure of a report that gets built over and over. And so just saying slap AI on top of it and hope it all works, like most data scientists, software developers will tell you, like, it's just too good to be true. Right? AI can certainly answer so many different kinds of questions, but it can't fix everything overnight. Right? You need to put the right structure in place to know that it's pulling from the right places, that you've tested it with different kinds of questions, and you feel confident on our answers. And, especially, it relates to a firm like ours that has a client aspect of the business. Right? We can't just have AI answer questions for us without reviewing what it's saying. Right? That we haven't battle tested it to give the right information to the right people. There needs to be a level of validation that a human is gonna have to play a role in it for the foreseeable future at least. And we gotta crawl before we walk, so we're trying to do small things first and see how they go and then build up into bigger things. And my advice is it's gonna take a long time for anyone to transform an organization in this way, so don't be impatient. I'm saying that for my job safety as well, but it's a long process.
[00:19:32] Breanna: I'd say the other final pieces that I would add is the people you task to lead this project are enormously important. So I'm gonna give Kunal kudos for being at the forefront here. And I'll also say that you have to have leadership buy in because as Kunal mentioned, it is a long process. And if leadership isn't fully on board, it's gonna stall at some point. And kudos to our team because it is coming from the top. And we've tried this in the past, and maybe it's petered out at some point. But right now, I think it is a top priority, and it's understood and has trickled down through the organization that this is a priority. And, therefore, I think the potential for success is much greater.
[00:20:12] Victoria: Incredible. Kunal, Bre, thank you so much for joining me. So many pearls of wisdom. Excited to share it with our audience, and thank you again for your time.
[00:20:21] Kunal: Absolutely. Thanks for having us.
[00:20:22] Breanna: Thanks for having us.
[00:20:25] Victoria: That's a wrap for this episode of the agentic allocator. If today's conversation gave you a clearer vision of where the industry is headed or helped you pinpoint exactly where your own process is stuck, go ahead and follow or subscribe wherever you get your podcasts and if you're curious what Agentic AI might actually look like inside your investment office and how to get there without compromising on security or control, visit auumai.com for demos and resources on AI-native LP and allocator workflows. Until next time.