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 Expect real-world case studies, tactical frameworks you can actually use, and moments that challenge outdated norms.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.
I think there will be a role for humans in the investment process. I think it's gonna be some time before AI methods are able to go to a company management board meeting and try to convince the CEO to take a new path of action that she may not have wanted to take.
Victoria Sienczewski:Welcome to The Agentic Allocator. Today, we're joined by Brad Calder, managing director at TIFF Investment Management, the OCIO serving endowments, foundations, and other mission driven organizations. Brad leads all aspects of marketable portfolios for TIFF and has spearheaded the firm's adoption and implementation of AI. How would you think about the value add component of AI within your investment and operational process over time?
Brad Calder:I think that's something like the tools we're building specifically for the investment committee that hopefully will make us make better decisions or make fewer decisions that result in large losses. I think if we can push those kinds of tools, that's where the true ROI will be felt the most.
Victoria Sienczewski:While many in the industry are just now starting to experiment, TIFF has been on the journey of AI adoption over two years. Today, we're discussing what AI adoption actually looks like on the ground, the friction of implementation, and Brad's vision for the end state of AI enabled investment organizations.
Brad Calder:Going back to your high school chemistry class, doing an experiment, you don't know what outcome of the experiment's gonna be. And I think a lot of these tools are very open ended in terms of how to use them. And so you have to try something and then tweak it and just not giving up. Because there aren't too many investment organizations that I think are worth backing that aren't investing in this technology pretty aggressively.
Victoria Sienczewski:I hope you enjoy the conversation. Brad, thank you so much for taking the time to join me for our conversation.
Brad Calder:Well, thanks for inviting me. It was my pleasure.
Victoria Sienczewski:Brad, TIFF is well along the path of AI adoption. When you look at your peers, you're honestly years ahead of LPs and allocators. Wanted to go back to the beginning on how you guys decided to really take the path of AI adoption, and and what kicked it off for
Brad Calder:you. So, we started thinking about this more carefully when we started backing machine learning oriented systematic hedge funds about 10 ago, and we learned about the benefits of the technology and some of the innovations that had happened with, you know, handwriting recognition or image recognition, and so we've been following this pretty closely, and once ChatGPT came out, we moved pretty quickly to try to find a provider that we could work with. And, you know, there's there's always just been a very innovative spirit here with respect to trying out new technology, and so it was a learning process. Some of what we started with isn't where we ended up, but, you know, that's been that's been the course of our journey, just to give you a sense.
Victoria Sienczewski:Amazing. And you mentioned the innovative spirit of TIFF. I witnessed it firsthand when I was an MBA intern writing an HVS case on TIFF. How do you really develop and foster that culture internally that allows the team to really embrace emerging technologies rather than be fearful of them?
Brad Calder:Yeah. I mean, one of the criteria that we use when we evaluate external investment managers, when we think about who we're gonna allocate capital to, a meaningful component of our decision is about how we see continuous improvement happening within their firm, and, you know, oftentimes, that's a a component of that is technology and thinking about how they could use technology to increase efficiency and make better decisions. And so we've just tried to apply that here within the team as well. I think also, we, you know, we have a a great leader on our tech team, Dave Snow, who's been at TIFF for over a decade. He came from Harvard Management Company before, and I think Dave's just been able to grow and retain a really great team of engineers, and that's very challenging in Boston, because it's a really competitive market for engineers.
Brad Calder:I think the fact that, you know, we've been able to build that kind of team, given that market backdrop, you know, says something about our culture and how we think about integrating technology into our whole our whole company.
Victoria Sienczewski:Incredible. And where did you start with AI adoption from a use case perspective? How did you map out kind of the highest points of friction or pain within your investment and operational process?
Brad Calder:Yeah. Like, the the big problem that we faced was we were using a RMS, research management system, that had been acquired by a a large company. I won't say which one because I don't wanna throw anyone under the bus, but we we felt like the company was not being adequately improved in a in a rapid fashion, and they, you know, lacked API capabilities so that we could run LOM models over that database, and we didn't really feel like there was a, like, an an urgency to improve. And so, you know, we we just felt like we had to seek other technology providers. I think that's that's something we're seeing pretty pretty carefully.
Brad Calder:I mean, this is happening now in the public markets with the SaaSpocalypse that we've seen over the last hundred days or so, where some of the companies that are perceived to not being at investing adequately are being, you know, punished pretty aggressively. Some of some of it might be over aggressive, but, you know, I I think that that's why there's such a white space for in certain verticals for AI companies to compete with the existing SaaS providers.
Victoria Sienczewski:Makes perfect sense. And would love to also dive into some of the use cases from an investment process perspective. Do you mind maybe walking us through one of the examples of a use case that you've been tackling with focusing on the ROI you've also been able to kind of generate from that use case?
Brad Calder:Yeah. I mean, so there's different use cases for different parts of the organization. So let's say you're just a individual group within the investment team, so the public equity team, when we receive information from prospective managers, we can have that quickly summarized in a template that helps us, you know, quickly vet an idea, think about ex expected return, and competitive advantage, and so forth. And we can also really quickly compare that to all of the investments that we currently have, so that we could quickly think about, hey, is this an upgrade, or or provide diversification, it's doing something different than we're already doing. We work with it to help us write investment memos, we have comparisons of quarterly letters, so you could see managers in quarterly letters over time, and you can talk to the matrix of that inform the entire information.
Brad Calder:Tell me about trends that you're seeing our managers in all investing around over the last three or four quarters. You know, what are what is what is only one of those managers doing that's different from the other, you know, you can ask these kinds of questions that, you know, historically, an analyst would have to read all those letters, have them all in their mind, and then try to think about patterns, I I think that's a challenging task, and certainly a time consuming one. Another thing we're trying to pilot that'll that's on our q two roadmap, for ex for example, is to have, like, a series of tools that aid the investment committee when making investment decisions. So our investment committee is composed of each of the asset class heads, so myself, the head of private equity, diversifying strategies, our portfolio head of risk, and our co CIOs, of course. And so, you know, there's varying level of tenure that each of these people have been at the organization.
Brad Calder:I've been here eleven years, you know, but some of them some of my colleagues have been here longer or shorter. So what we're we're planning to build is, you know, have investment memos from every investment TIFF's ever made going back to inception, and then, you know, we have once we can tie that to the structured returns data and exposure data that we we are we have, but it's not merged just yet, but the goal would be for us to have, like, an automated report that says, hey, this investment is most similar to these other three investments. Here here were the outcomes. You know, I think that gives us a better way to sort of interact with the full investment history of our company. Our company is 35 years old, so no one's been here since the inception.
Brad Calder:But we're so I I think at this point, we're not, like, adequately learning from the collective wisdom from the investments that were made prior to any of us being on the committee, and so there's there's all sorts of tools that we're kind of building for that are, like, specific to each part of the investment process, so I hope I hope that's helpful context.
Victoria Sienczewski:Absolutely. What a fantastic use case. I love the approach to really defining TIFF's collective wisdom as an organization or that kind of collective intelligence layer. Phenomenal. Would love to also dive into the use case you mentioned earlier from a screening perspective.
Victoria Sienczewski:One thing that I've been thinking about in the kind of OMAI context has been around defining an OMAI score for a manager. I'm curious if TIFF has been thinking about that as well from a screening perspective. I know you have a very kind of clearly defined framework for how you evaluate managers, but curious if that's also part of what you're considering.
Brad Calder:Yeah. We are experimenting with that. It's not something that we've been super thrilled with the output right now. It's getting better. You know, we we have a manager ranking process that we conduct annually, where, you know, there's, you know, five central categories and there's subcategories, and we rank on every subcategory and broader category on a one through four scale, so we fed that into our our model, you know, we started doing that rigorously since 2017, and kinda have to teach it over time, and I think one of the challenges there, and this is, you know, something that, you know, where I I hope we can have some some more data science help over time.
Brad Calder:We're probably gonna we're thinking about what we should do on this front, but, you know, you wanna make sure that the the data that you have is, like, very clearly clearly time stamped just to avoid like look ahead bias, if you're gonna start using those kind of scoring rubrics that a lot of people have for either like forecasting purposes, or for just evaluation purposes. So I think there's a lot of promise there, but we're not there yet. We're just kind of still in the experimental phase on that.
Victoria Sienczewski:Fantastic. And curious, how are you defining ROI internally for your AI projects?
Brad Calder:Well, I mean, the initial ROI is if we can get the current tool to a point where we can get rid of our existing tool, then at least it's cost neutral, and we're not there yet. So I do believe we'll be able to get there this year. And after that, I think it will be a little bit harder to estimate. I mean, we're not thinking about this necessarily as, hey, can we reduce headcount by using more agentic actions? You know, we have an offshore team in India that helps us.
Brad Calder:We've we've found that, you know, combining BPO and AI tools, and our team there has full access to all of our AI tools can be very complementary, as opposed to some sort of, like, substitute for people. But I I would expect us to be more judicious when thinking about expanding the team domestically. We can do a lot more with the same number of people, but I'm not sure it's gonna lead to some sort of like widespread force reduction.
Victoria Sienczewski:Absolutely makes sense. And curious, Brad, in terms of you mentioned ROI focused on efficiency, productivity gains. How do you think about in the future, and I know that you're on a journey, right? The entire industry is on a journey, but how would you think about kind of the value add component of AI within your investment and operational process over time?
Brad Calder:Yeah. I mean, I think that's like, something like the the tools we're building specifically for the investment committee that hopefully will make us make better decisions, or make fewer decisions that result in large losses, or, you know, I I think if we can push that kind of those kinds of tools, that's that's where the true ROI will be felt the most. And, you know, I think for some other instances, particularly on the private side, where speed of execution is extremely important, I think we're, you know, experimenting with AI oriented legal tools as well. The results so far have been mixed. So, you know, we still retain external counsel, like traditional external fund counsel as opposed to exclusively using AI oriented legal services, but we are, if for certain types of legal work, willing to go with an AI oriented tool that's much less less costly.
Brad Calder:So I I think that one's still more of a a cost benefit or risk downside risk consideration may drive those may drive us to, you know, favor, you know, human oriented counsel, at least for now. But, you know, we're we're experimenting with that too.
Victoria Sienczewski:How have you thought about the the challenges of AI adoption? We've discussed a lot of the benefits and hopefully the forward looking additions to the process, but curious, what challenges have you encountered along the journey?
Brad Calder:Because the foundational models keep improving, you know, you could come to the conclusion that a project that you experimented on once or twice that didn't work, you might draw the conclusion it's can't work, period. I think that could be an incorrect conclusion, and we've seen the we've seen enough improvements on that that, you know, I think it's worth continuously experimenting and trying things and seeing, hey, maybe this new version of whatever Claude, you know, Gemini Chat GPG, whatever, could help you more than, you know, the previous version six to twelve months ago. But also there's just like user comfort as well, and I think for for some users, like, if they you know, someone on our client team might have a quick question about an investment we've made. You know, they make a query, they click the you know, they're reading it, but they're not really super confident in the output. Even with the validation, you know, you it they still might wanna follow-up or maybe they don't fully trust the output.
Brad Calder:And and so I think there's some kind of user comfort in having people kind of figure out how to self validate data that takes some time. But I I I think there's a widespread view at TIFF that this technology is broadly helpful. Like, I don't think there's too many big skeptics or folks that are just like, you know, not willing to at least try using our systems for some purposes, but it it's it's it's not like, there is variance on that on that point across the company.
Victoria Sienczewski:Brad, and when you look, let's say, three, five years down the road, maybe even ten, what do you think an AI enabled LP or allocator will look like at that point?
Brad Calder:So I I think there will be a role for humans in the investment process for some time to come. There's a lot of things that we do though that are high highly likely to be done better by AI tools, and, you know, when you think about part of the investment process, you know, you're taking in a lot of data, and at some level, implicitly or even explicitly, we try to be explicit here about this, but implicitly, you you are making a forecast about what's gonna happen in the future. I I would expect TIFF to be hiring people with explicit expertise at taking textual data, and then making forecasts about it, and if you could tie in accurate numerical data for that, you might be able to take in data from a a potential manager, and there would be a probability instantaneously provided probability about their expected outcomes that could come from a strategy like that based on the description and or other data that we have about similar types of managers. And I would expect AI to help us propose due diligence questions that it believes are more predictive of forward outcomes. Some of those may be very reasonable, some of them may be, like, hard to understand and why that's relevant, but I I could imagine in the future, you know, the the entire due diligence process may be done with more input from, like, agenda creation and what's what's truly relevant from our a AI tools making recommendations.
Brad Calder:I believe at least that, you know, the human side of it where, you know, we're negotiating on terms or, you know, with our external GPs when they're taking meetings with company management, etcetera, you know, I I think it's gonna be some time before, you know, AI methods are able to, you know, go to a go to a company management board meeting and try to convince the CEO to take a new path of action that she may not have wanted to take. You know, I think the the, you know, the control rights that, you know, owner business owners have when they own equity of public or private companies, etcetera, I think those control rights are still gonna be hopefully leveraged by humans and not delegated to AI systems. It's a little scary world if that that that comes to pass. So I I I think that's where, you know, we should be really digging in, at least on the investment side, you know, thinking about investors that can, you know, leverage their control rights in order to push for corporate actions that are favorable for, you know, all stakeholders, but, you know, hopefully for, you know, the equity owners as well.
Brad Calder:Just because I think that's gonna be an area that's harder to disintermediate.
Victoria Sienczewski:A fascinating future for sure and I agree with you. I view the integration of AI as freeing up teams to focus on the human component as you mentioned, that there really isn't a substitute for, and we shouldn't have a substitute for. Another area that I'm really interested in kind of in the future is what this interface of what I call like an Agentic Allocator will look like. Right now in many cases folks and firms are focused on trying to integrate it into a platform that they can log in and run a workflow or run an agentic orchestration step by step, but over time I think that these systems are going to be much more integrated into firms existing tech stacks and you'll maybe not even need to query it, it'll just know what's happening and know all of the context. So still to be seen, but really, really exciting.
Brad Calder:Yeah, absolutely. And I I I think that's probably not too far away. I think that's that's that's much close. I mean, assuming that you wanna make the investments to to do that, and I do think it will become a little bit, and it already is, but of course, but I think it will continue to be a scale game, You know, like, for a really subscale endowment and foundation allocators that, you know, lack the budget to make these kinds of investments, I do wonder the you know, how that's gonna impact them, and, you know, that that's part of why, like, TIFF exists, to, you know, help smaller charities, you know, solve for these kinds of problems in the same way that, you know, our larger, you know, board member organizations are solving for them, and we're, you know, we're trying to learn from our board members as well here at TIFF. I I do think, though, it's it's not gonna be, like, a a rising tide that lifts all ships.
Brad Calder:I mean, it's only gonna lift you up if you're willing to greatly make the investments to kinda make this happen, in my view.
Victoria Sienczewski:Super fascinating. And Brad, final question to wrap up. For your peers that are, you know, not as far along within the AI adoption journey as you are, curious what first step would you suggest or what what approach would you suggest they take when exploring these technologies?
Brad Calder:I think just, you know, going back to your high school chemistry class, right, you're you're in in a position where you're doing an experiment, and, you know, you don't know what the outcome of the experiment's gonna be, and if it doesn't work, you alter the experiment a bit and try something new, and, you know, I think a lot of these tools are, you know, it's very open ended in terms of how to use them, and so you have to try something and then tweak it if it's not working, and just not giving up. Because it's it's very easy to just feel like, oh, it's a lot of work, like, to do experiments to improve, but, you know, there aren't too many investment organizations that I think are worth backing that, you know, aren't investing in this technology pretty aggressively. And so, yeah, I I think it's it's worth it, and and, you know, it's like going to the gym. Like, if you're not gonna try, like, you're probably not gonna get much stronger.
Victoria Sienczewski:Wonderful. Brad, thank you so much for taking the time to join me. Incredible pearls of wisdom.
Brad Calder:Yeah. Happy to help and, you know, love to keep in touch on these issues and, you know, learning from you too. So thank you for the invite.
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. 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.