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 Expect real-world case studies, tactical frameworks you can actually use, and moments that challenge outdated norms.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.
The core issue with LLMs and the way they think is that the model is baked, it's fixed, it has no memory, and it's not pliable. That's why you need to build a brain. And their context layer, what they can have access to for any given task, is incredibly important. And the more relevant context you can give them, the better their output is going to be.
Victoria Sienczewski:Welcome to The Agentic Allocator. Today, we're joined by Richard Perris, co founder and president of Covenant, an AI enabled law firm built for investors in private markets. Richard brings more than twenty years of experience and he spent fifteen years as general counsel of CDC, leading legal and compliance across its private equity and private credit businesses.
Richard Perris:When I first started actually law back in the early two thousands, there were still people who would say, you'd be nuts to send client documents by email. And people don't say that anymore. People would come to a landing on what is kind of a reasonable level of risk around it. Security. I think that's starting to happen with AI now a little more.
Victoria Sienczewski:Covenant's DNA is that of a law firm. But it uses AI to give LPs a clearer view of the key terms across their fund documents so they can negotiate with better information. The firm has moved away from billable hours entirely and is building towards a model where its legal advice and data plug into the AI systems its clients already use.
Richard Perris:I think interconnectivity is actually going to allow us to do things that weren't really feasible in the past. So all this brain is storing all this information about an institutional investor much more effectively than any human brain could ever store the same information. You know, an AI doing a legal review can be given knowledge of that database, say, oh, well, here's a point that's come up. It will be interesting to know, you know, what the last five things they did about that were. I'll go find that.
Victoria Sienczewski:Today, we discuss how attitudes towards data security in the context of AI have shifted, how AI data and legal advice are setting a new balance between LPs and GPs, and what the future holds for private market investors. I hope you enjoy the conversation. Richard, thank you so much for taking the time to speak with me today.
Richard Perris:It's an absolute pleasure.
Victoria Sienczewski:You spent fifteen years as CBC's general counsel. Through that experience, you spent hundreds of hours, yes, hundreds of different negotiations on the GP side. What did that experience tell you about how you work with LPs today?
Richard Perris:So GP LP negotiations have always had a sort of weird disparity to them. There's a one sidedness to them. Private markets have been very popular with investors over the last couple of decades. So as a GP, if you're any good, your funds are normally oversubscribed. You also have the comfort that investors are paying everyone's legal bills.
Richard Perris:So on the legal side, the negotiations, you know, always did feel quite straightforward from the point of view of the manager. So there is this, you know, this disparity has existed in the market for a while. I don't think it's been a particularly, you know, healthy, you know, drift for the market. GPs have been able to to command quite favorable legal terms. So when I left CBC after COVID turned the world upside down, you know, the one thing I wanted to explore was is there a better way, a better solution for LPs that can sort of help them in this in this market?
Richard Perris:And it coincided with the emergence of the technology to do it. So it was like, you know, someone had given me a gift to go try and see if we could make, you know, legal review for LPs a little bit more effective, a little bit more cost effective, more efficient, and frankly, quality. That's what we set out to do at Covenant.
Victoria Sienczewski:Amazing. A couple years ago, right, think back to 2023, the start of every conversation with LPs was around data privacy, data security in the context of AI. Are conversations still starting focused on data privacy and security or has that evolved?
Richard Perris:There's definitely been a shift, you know, really in the whole conversation about AI as everyone knows, you know, over the last two or three years. When we got started, it was still a relatively new technology in everyone's minds. So, you know, everyone was rightly questioning it and, you know, asking questions about, well, how does it work? The technology arrived so rapidly that most people hadn't had a chance to kind of fully absorb the implications of it from any real angle. And IT security was sort of top of everyone's list.
Richard Perris:And, you know, there wasn't this, imperative to kind of use it that has developed over the last two years. Now, you know, the the conversations we had in 2023 where we had to go to great lengths, you know, to protect people, make sure we're protecting people's data and, you know, it took a long time to convince people to even give it a go. Now the kind of the attitude is shifting and, the, you know, everyone now starts a conversation with, okay, I want to use AI, you know, that's my starting point. You know, are you offering a solution that can help me, you know, deploy this new technology to my benefit? And then, you know, that that obviously helps.
Richard Perris:You know, the data security questions are still there, of course. But now people are just very keen to find a solution rather than, you know, raise an issue. So the market is definitely starting to understand the technology better. People are starting to get more comfortable with it. Just like any technology, like when I first started practicing law back in the early two thousands, email was still relatively new.
Richard Perris:And people would... There were still people who would say, you'd be nuts to send client documents by email. Right? Why are you doing that? That you didn't...
Richard Perris:Don't you know that that's, you know, on the Internet everything can just be seen by everyone? And people don't say that anymore. You know, people have sort of come to a landing on what is kind of a reasonable level of risk around data security. I think that's starting to happen with AI now a little more. And, you know, there there are just certain things you have to do if you're if you're deploying it in an enterprise context.
Richard Perris:And if you do those, then you kinda get to it because now there's this imperative. People are seeing what it can do and, yeah, they wanna deploy it in the real world and benefit from it.
Victoria Sienczewski:Phenomenal. I've seen that shift as well. We're a much younger company, but in terms of the conversations and how they've evolved around, of course the risk remains, yes, but how do you think about that risk? What are you comfortable with? How do you put in place the guardrails or the governance structure to make sure that the team is using it?
Victoria Sienczewski:Because the other element that I found is that if you don't have a policy in place or you don't have governance in place, well, your team is probably still using AI.
Richard Perris:The whole issue of shadowing... Exactly. So, you know, it's such an irresistibly useful and obviously useful technology. You know, if you're not using it, you're clearly missing out. So, you know, that's why, you know, people find a way.
Richard Perris:You know, that they they find a way to to to make it work.
Victoria Sienczewski:Covenant has a very interesting business model. You actually shifted away from the concept, right, or the framework of billable hours. Why did you make that change?
Richard Perris:Well, the billable hour has always been sort of an awkward compromise when it comes to a means of valuing knowledge work of any kind, but especially legal services. Know, they're so, you know, weighing things by the hours that they took individuals to do is kind of an odd way of valuing the output, if that makes sense. You know, the reason it's persisted is because it's hard to value the output, frankly. You know, a lot of legal services, you know, aren't necessarily directly linked to, you know, an economic outcome. So, you know, and do take time to do by well educated individuals who deserve compensation for that time.
Richard Perris:Right? So it's a natural way to to sort of... As a starting point for charging, but it does put the incentives in completely the wrong place. Right? So, you know, you...
Richard Perris:Who would... You know, if the quality stays the same, who would wanna pay more if work takes longer to do? You know, from the client's perspective, they want to incentivize efficiency. So it's always been a little bit of an awkward way of valuing legal work in the first place, and the way it's been dealt with is just by negotiation at the end of every transaction. Like, you add up all the hours and you say, okay, it's a rough proxy for how complicated and difficult it was, right, but that's only ever been a starting point, and it entirely relies on it being true that legal work takes time by humans.
Richard Perris:AI arrives, completely destroys that business model. AI is capable of doing most tasks that associates and law firms currently spend their time doing, and the law firms need them to spend time doing because that's the only way they can actually charge enough of the task to give the partner the compensation that's necessary. So you once AI arrives, you really have to start thinking about changing that business model. And that's kind of one of the the catalysts that meant we would we would start a new law firm as covenant rather than sort of going getting a job because it's easier to build it from the ground up around the technology than it is, you know, to sort of shift a cruise ship around a corner and turn around, which is the challenge that major law firms are currently dealing with. You know, how do we reorganize the business around this technology?
Victoria Sienczewski:Fascinating evolution. Yeah. As you know, another interesting evolution is the ecosystem. So when I look at an LP's tech stack, how that stack looks like is going to rapidly evolve. And it already has in many cases for forward leading LPs.
Victoria Sienczewski:And so what I'd love to dive into at OMAI we're building the AI brain for investment management. And what I'm interested in understanding getting your perspective on is how does an AI brain that OMAI is building interact with covenant and the incredible legal data that you have?
Richard Perris:Well, I mean, the fact... One of the fascinating things about this technology is how easy it is now to integrate different sources of data. You know, because you can now apply intelligence to almost anything, you know, without the need for expensive, squidgy, sleepy humans to figure it out, you can, you know, you can link things together in a much more flexible way. And you really need to do that if you're gonna then deploy the technology to do other tasks because, you know, the the core issue with LLMs and the way they think is that they, you know, the model is baked. It's fixed.
Richard Perris:It has no memory and it's not pliable. It can't learn anything more in any real sense like our brains, you know, changing every single second of every day. You know, our mental neural network is, you know, is very pliable. And that's, I guess, what we call memory. LLMs don't have that.
Richard Perris:They they they can only learn from the context that they're given, you know, anything that fits within their context window can change how they react to something else, but otherwise they can't remember things. That's why you need to build a brain, And their their context layer, what they can have access to, you know, for any given task is incredibly important. And the more relevant context you can give them, you know, the better their output is going to be. So, you know, yeah, organizing the context layer and giving, you know, the AI, you know, model or agent or whatever you want to call it, access to the right context at the right time and the ability to access that at the right time is gonna become an incredibly important part of life and the ecosystem going forward. And I think interconnectivity is actually going to allow us to do things that probably weren't really feasible in the past.
Richard Perris:So, you know, if we're giving a client a piece of legal advice, you know, we're doing a review of a document, the possibility exists. Well, the AI that's doing that review could decide that something might be relevant and immediately go and get it from somewhere. So, Ohm's, you know, brain, you know, is storing out all this information about, you know, an institutional investor, probably much more effectively than any human brain could ever store the same information. You know, an AI doing a legal review, you know, can be given knowledge of that database, you know, in a structured way and say, oh, well, here's a point that's come up. It will be interesting to know, you know, what the last five things they did about that were and I'll go find that.
Richard Perris:Right? And and, you know, we have that data from a legal perspective on our system, but legal data isn't necessarily the end of the story. So, you know, I think interconnectivity between all these different data pools is gonna become a very interesting feature. And allowing clients to access data where they are, you know, where they're actually working is gonna be very important. So for example, you know, one thing we're currently working on is an MCP...
Richard Perris:So MCP connector to allow our clients to access Covenant from Claude or Chateapiti. Right? Because that's where there were... A lot of them were working in Claude. Why should they have to switch platforms to ask a question about an LPA review that, you know, or or some of the legal data that we have stored.
Richard Perris:We just want them to be able to tag Covenant and say, okay, Covenant, get me this information, please. And that means they can give... If all is also available to Claw at that time, you know, then that's where the combination can happen and the magic can happen to kind of help them achieve something that previously probably would have taken a lot longer with a lot more human and a lot more phone calls and a lot more emails, and now it's all just right there in Claude. It's... The possibilities are pretty phenomenal.
Victoria Sienczewski:I share your optimism and excitement for the future in terms of interconnect ivity. And I see, of course, the importance of the context layer, right, that institutional knowledge accessible from a legal perspective or from an operational or investment perspective, but also interconnectivity among different teams within an organization. So I love it. Right? We've talked about how the broader ecosystem connects and what that'll fuel internally within organizations.
Victoria Sienczewski:How we break down those knowledge silos between a historical investment research team and a client facing team, for example.
Richard Perris:Yeah. Yeah.
Victoria Sienczewski:It's fascinating.
Richard Perris:Well, I mean, the private market is such a great use case for all of this because it's like, you know, I come over from New York to The UK, and I need to remember to bring an adapter for the plugs. Right? You know, otherwise, I've say this because I forgot my adapter. Yeah. Yeah.
Richard Perris:It's a pain. I... Why... I'd... I still don't know to this day how that happened.
Richard Perris:I mean, I guess, you know, just travel wasn't as common, but there's literally, you know, then you go to Switzerland, it's slightly different to the one that's in Germany or whatever. Alright. Private markets is very like that. You know, one fund, one SMA, you know, use a different template, a different format, know, different... It's it's all the same stuff, but it exists in very different, you know, formats and they can't plug together very easily.
Richard Perris:In order to make them plug together previously, someone would have had to do that work pretty manually, you know, going through them and putting the database together in like an ordered way. Now that someone doesn't have to be a someone, it can be an AI. And that's kind of an exciting thing that suddenly, you know, the interconnectivity that sort of slowed down private markets or the lack of interconnectivity or or compatibility that slowed down private markets, you know, will disintegrate, because, you know, suddenly it's cost effective to allow allow it all to kind of just be connected.
Victoria Sienczewski:And the opacity, right, of the private market ecosystem. What I've also seen is for example like portal providers. So it blows my mind that we can have AI do these amazing things and there are still ingrained portal providers where it is again, a huge security risk to unleash Muse or Openclaw to try to download your files and there's still no API or MCP connectors. That'll have to change.
Richard Perris:I think I think it'll it'll become a, you know, it'll be like table stakes. Like, the people aren't gonna tolerate, you know, having... In in an environment where they know because other people have done it, that everything can legit, you know, be accessible through one platform that they're already using. They're not gonna tolerate, you know, having to switch.
Victoria Sienczewski:In a world where LPs have more access to their data, right, an overview of all of their historical legal negotiations and terms, right, through through covenant, for example. And GPs are also adopting their own suite of solutions, which empowers them. How does that change legal negotiations on the fund side?
Richard Perris:That's an interesting question. I mean, ultimately, market dynamics are still gonna play a huge part. Right? So the, you know, the oversubscription to the fund is still gonna allow GPs to say no on a lot of terms. But frankly, the industry has been crying out for standardization for a long time, and I'm really optimistic that, you know, things are gonna get a little easier to standardize because you'll you'll have that visibility.
Richard Perris:Most of the terms in limited partnership agreement deal with edge cases. You know, stuff that never really comes up. You know, it's one in 500 funds that kind of go over their term and then have litigation and the GP gets sued and has to claim under the indemnity five years after the fund was terminated. Like, that's a very... That's an edge game.
Richard Perris:It happens and in the fund it happens. The wording is very, important. But that was wording that was negotiated, you know, fifteen years before, you know, really we should have standard work. And so everyone knows what would happen in that edge case and you don't really need to try and worry about every single fund you invest in because you know that for 99.99% of them, it's never gonna be, you know, red again. So I'm really hoping that because you can then, you know, demonstrably say, okay, this is the standard, this is the market, you know, we like to show our clients their own market.
Richard Perris:Know, part of our platform is really showing you, okay, what have you agreed to in the past? Right? So, you know, you can look back over the last 200 funds you did and you can say, okay, this indemnity clause you put in front of me is very, very different from every other one I've ever agreed to. Why did you do that? Right?
Richard Perris:So it really empowers you to kind of like sensibly make those points with backup because previously to make that point I would have had to call a very expensive law firm and say, Could you look back over the last 200 LPAs that I agreed to and just compare them to this? And they would go, Yes, we'd love to do that. Put the phone down. They would come back probably a month later in that case and say, we've done it. It's very, very different.
Richard Perris:You shouldn't agree to it. And the bill's on the way and you're like, oh, that's a 250,000. It probably wasn't worth a 250,000 to negotiate this edge case clause that probably would never come up. Right? That's gone as a problem now.
Richard Perris:It doesn't exist anymore. You can go onto covenant, you can go look at your data, and you can say it by way, you know, two minutes and you're done. So that, yes, it will have an influence on stuff like that, I'm sure. Obviously, on the core economic terms, I think, you know, the market forces are still, you know, gonna be the the dictating force, but I think getting rid of the pain of the rest of the detail is gonna be the first thing that happens. But that's the optimistic case.
Victoria Sienczewski:Would you like to share your pessimistic case? You're you're leading with that, Richard. Human
Richard Perris:beings do things in weird ways sometimes. So, we don't always act, you know, in our common good. And if technology is capable of, like, doing more work more quickly, that can almost create work. Back when I started, for example, I got a chance again in the early two thousands when people were questioning email, they were also sometimes sending handwritten markups of drafts through the post to the other side, right, rather than sending them by email. It was like a traveling...
Richard Perris:Now, when you're doing that, you take very good care that you're very sure you want to make the comment you're making because it's so much hassle to do it, and then the person receiving it, again, will make very good take very good care that they care about it enough to kind of respond and put something else in. Right? And, like, you're fiddling around with it forever. Word processing and email completely eliminated that issue, but contracts got a lot longer. Like, you know, when I...
Richard Perris:I I actually dug out when I was at CBC the very original contracts for CBC's spin out from Citigroup, and I was like, oh, this is interesting because it's a historical artifact. But the thing that struck me when I read through them was how short they were and how to the point and concise. Right? Like the original sort of... Even the fund agreement, you know, was far far shorter than than any LPA you'd see today, which are 100 pages long.
Richard Perris:Many people would blame the technology for that because it makes it possible, you know, to put all that complexity into the document really easily, right, and to sort of negotiate around it. There's a danger that AI does exactly the same thing, right? Because you can do more, you know, drafting and, you know, faffing about, you do it because, you know, why not? Right? Because the...
Richard Perris:And so there is a world where it actually proliferates complexity. That's the pessimistic case. But I don't think that'll happen and, yeah, the... It's gonna be just... It's very hard to say.
Richard Perris:Like, AI could be used to then counteract that and maybe it doesn't matter. Maybe it... Maybe it's...
Victoria Sienczewski:Have an agent write it. You have an agent review it.
Richard Perris:Everything's hyper efficient. They all go into detail. You know, it's just points that should have been covered, but, you know, we never got around to. So, you know, it's a fascinating... But, you know, I don't wanna make predictions.
Richard Perris:I just made two predictions, but they're opposite predictions, so I'm hedging.
Victoria Sienczewski:Exactly. And we'll and we'll see which one proves true.
Richard Perris:Yeah. Exactly.
Victoria Sienczewski:Richard, to wrap things up, would love to ask you for your advice. So if you were to speak with an LP or a GP earlier on in their adoption journey of AI, what would you tell them? Where should they start?
Richard Perris:In their adoption journey? Well, if you mean in their own use of AI, I mean, there, I... My strong advice would be to start using it immediately, and, you know, use it for as much as you can possibly come up with. Be ambitious about the use cases that you're gonna deploy it for because frankly, get yourself a Claude Cowork subscription and start asking it to do things and write skills to allow it to do things and, you know, you you'd probably be quite... If you're assuming this is...
Richard Perris:I don't know if there's anyone out there who hasn't used Claude or or Chagiboutin, they probably are, but you'd probably be absolutely astounded at what it can do, you know, just given a little bit of of guidance about what you want. That would be my very strong advice to someone. In terms of service providers and, you know, what you should be demanding from them now, you should be asking them how... And, you know, I think clients ask that in, you know, in the legal context at least, you know, how are you deploying this technology? How is it gonna benefit me?
Richard Perris:Are we gonna have a discussion about pricing? Are we gonna have a discussion about value? Those kind of conversations should be happening now, and then the expectations rightly should have shifted in terms of, you know, your lawyers, you know, deploying AI in an efficient way. So I would definitely advise people to have those conversations as soon as they can as well.
Victoria Sienczewski:Richard, thank you so much for taking the time to speak with me. It was a fascinating conversation.
Richard Perris:Well, it was a pleasure. Thank you very much, Richard. Thank you.
Victoria Sienczewski: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. Your firm's judgment took years to build. Stop letting it reset every time a team member leaves. Visit auumai.com to discover the AI brain your team needs to leverage AI to its fullest potential.
Victoria Sienczewski:Until next time.