
Sign up to save your podcasts
Or


Al Hemmingsen, CIO of a prominent US-based single family office, joins The Agentic Allocator to share what hands-on AI adoption looks like from inside an institutional family office. Al is the only CIO, we know of, in the LP world actively publishing open-source code on GitHub for use by fellow allocators.
In this episode, Al walks through how vibe coding with Claude Code became a serious tool for unlocking legacy data and automating workflows, how he is building a culture of accountability across his team, and why the LP that gets to the best GPs first has a competitive edge that only widens over time.
He also makes one of the sharpest points we have heard on the competitive stakes: some aspects of this work should be shared freely with peers, but the ability to process information faster than other LPs, build higher-quality manager meeting lists, and get to the best GPs before they fill their calendars at conferences, is a genuine edge. If you do not build it, someone else will.
What You'll Learn:
About Al Hemmingsen:
Al Hemmingsen is the CIO of a prominent US-based single family office with a history going back to the mid-20th century. He has been at the helm for the past decade, managing a multi-generational pool of capital with a quasi-perpetual mandate. Al is one of the most hands-on AI adopters in the institutional family office world and is the only CIO, that we’re aware, of actively publishing open-source code on GitHub for use by fellow allocators.
Episode Highlights:
[05:56] The $59.32 Moment
$59.32 in API credits unlocked 50-plus years of legacy transaction data in a single afternoon. That moment set off a systematic search for every legacy problem in the organisation AI could solve.
[09:18] Building in the Seams
What is worth building are the tools that live between enterprise systems: workflows too specific for any software company to build a product for.
[11:23] Taking Fear Off the Table
The goal is to eliminate the 30% of every job that is tedious and repetitive, not the people doing it. Remove that fear first and the rest follows.
[19:09] The Investment Case
There are 56,000 alternative managers globally. The LP that processes information faster and gets to the best GPs have a genuine edge.
[20:55] Collaboration and Competitive Advantage
Al's view is that sharing tools and ideas freely with peers in the family office world makes everyone better, which is why he publishes on GitHub. But that openness has a natural boundary: the capability to see more managers, filter faster, and reach the best GPs before their schedules fill is a genuine investment edge worth building and protecting.
Episode Resources:
Disclaimer: The views and opinions expressed by Al Hemmingsen in this interview are his own and do not necessarily reflect the views, positions, or policies of any organization with which he is or has been affiliated.
Professor Ludovic Phalippou of Oxford Saïd Business School joins The Agentic Allocator to deliver a rigorous, unsparing look at what AI can and cannot do for Limited Partners in private markets. Drawing on his research paper, 'Limited Partner Versus Unlimited Technologies,' and hands-on experiments applying machine learning to real LP datasets, Professor Phalippou maps both the transformative potential and the structural risks of AI adoption in private markets.
Private markets are not data-poor - they are data-overwhelmed. A single fund investment can generate thousands of pages of PDFs and countless spreadsheets. A large LP like CalPERS monitors 200 active private equity funds and receives quarterly reports on 3,000 underlying portfolio companies. No human can absorb that. The promise of AI is that it can. But Phalippou warns that the same tools capable of unlocking hidden signals in qualitative data can just as easily be gamed, misused, or deployed in ways that amplify the industry's existing distortions.
Professor Phalippou shares research showing how AI sentiment analysis of GP quarterly reports can predict future portfolio company returns with real predictive power, far beyond what the reported marks alone convey. He also lays out in precise detail how GPs could exploit AI-reliant LPs, from burying critical fee terms ever deeper in footnotes to inserting invisible prompt injections into documents. He outlines the governance framework LPs need to build before they trust any AI output with a real decision.
What You'll Learn:
About Professor Ludovic Phalippou:
Ludovic Phalippou is a Professor of Financial Economics at Oxford’s Saïd Business School and specializes in Asset Management, with a special focus on investments in private equity funds. He is one of the world's leading academic authorities on private equity performance, fees, and transparency. His research on private markets has been widely cited by LPs, regulators, and policymakers. He sits on the investment committee of Oxford’s Queen’s College, has collaborated with institutional LPs on applied machine learning research, and authored the paper 'Limited Partner Versus Unlimited Technologies,' which examines how generative AI and machine learning are poised to reshape LP decision-making, and the risks that come with it.
Episode Highlights:
[00:03:21] The Real Data Problem in Private Markets
Phalippou dismantles the assumption that private markets suffer from a lack of data. The problem, he argues, is the opposite: thousands of pages per fund, quarterly reports on thousands of portfolio companies, and none of it structured or comparable. AI's promise is to swallow that volume and surface what matters.
[00:04:31] Sentiment Analysis as a Return Predictor
Drawing on original research, Phalippou explains how running AI sentiment analysis on GP quarterly report text produces a powerful predictor of future portfolio company returns - one that adds significant information beyond the reported mark. Two companies both marked at 1.2x can have dramatically different outcomes depending on how positively or neutrally the accompanying text is written.
[00:09:19] Why EBITDA Extraction Is the Dangerous Use Case
The most common instinct for LPs, using AI to automatically extract and aggregate financial metrics across hundreds of portfolio companies, is also the most dangerous. EBITDA figures in private markets are loaded with definitional choices, add backs, and adjustments buried in footnotes. Asking AI for the headline number without the context around it can produce false confidence.
[00:11:40] How GPs Can Game AI-Reliant LPs
Once GPs know their documents are being processed by AI rather than read by humans, the incentive structure shifts. Critical fee terms and performance qualifications will be buried even deeper in footnotes. More alarmingly, Phalippou explains how invisible prompt injections, text hidden in white-on-white font, can manipulate AI summaries to generate favourable assessments. Academics have already been caught doing this with peer review systems.
[00:13:29] What Good AI Prompting Looks Like for LPs
Rather than asking AI for a single extracted number, Phalippou recommends asking it to surface all relevant passages in a document on a given topic, including definitions, footnotes, and any conflicting disclosures. The absence of a clear definition is itself an information signal. Over time, analysts can train AI on their own judgment: flagging specific types of add backs, identifying patterns that have historically indicated problems.
[00:19:29] The Legal Frontier: AI Provisions in LP/GP Agreements
Some GPs are already attempting to prohibit LPs from processing their documents with AI. Phalippou argues the logic does not hold: when an LP hired Cambridge Associates, they were paying for judgments informed by thousands of comparable documents. Giving a document to an AI does the same thing. The conversation predicts LP/GP confidentiality agreements will soon include explicit AI provisions, a structural shift in how the industry governs information.
[00:21:00] The Machine-to-Machine Future
Phalippou reflects on an emerging world in which GPs produce documents with AI assistance, LPs process them with AI, and the analysis in between is conducted by AI. He uses the analogy of education: when exams are written, answered, and graded by machine, what is the human layer actually doing? His answer: setting the values, defining the questions worth asking, and retaining genuine accountability for the answers.
[00:23:29] A Governance Blueprint for LP Organisations
Phalippou's closing framework centres on responsibility. When an AI-extracted number turns out to be wrong and a real investment decision follows from it, who is accountable? Organisations must resolve this before deploying AI, not after. He also stresses the importance of avoiding both naive adoption ('this will simplify everything') and reflexive dismissal ('it just generates noise'). The mature position is to use these tools seriously, with clear governance and a clear-eyed understanding of the risks.
Episode Resources:
Professor Ludovic Phalippou on LinkedIn
Oxford Saïd B...
Marine Eugene, Chief Commercial Officer at Moonfare, explains how one of the private markets' most recognised digital platforms is now embedding AI across its investment process, commercial operations, and client experience, without compromising on the high-touch service that sophisticated family offices and investors demand.
Moonfare has built its reputation democratising access to top-tier private market funds by removing friction from the investment process. Now, with a team of 15 investment professionals monitoring over 250+ alternative funds annually and a client base spanning 23 markets, Marine explains how Moonfare is deploying AI to handle the scale of data processing, lead triage, and compliance monitoring that simply wasn’t possible manually.
In this episode, Marine traces the full arc: from the early days of digitising paperwork and onboarding flows, to deploying AI for legal document extraction, Excel automation, and IC memo generation, and into the next wave - using AI to surface fund managers, track live portfolio performance, and deliver the kind of interactive, real-time data experience that family offices are now expecting as standard.
What You’ll Learn:
About Marine Eugene:
Marine Eugene is Chief Commercial Officer at Moonfare, the digital private markets platform recognised for democratising access to institutional-quality funds. Prior to Moonfare, Marine was an executive in the private aviation industry, including at NetJets and FlexJet. Over the past three years at Moonfare, she has led the firm’s commercial growth across 23 markets, with a focus on serving family offices and sophisticated institutional investors through a high-quality, digitally-native experience.
Episode Highlights:
[00:01:36] From Private Aviation to Private Markets: The Digital-First Shift
Marine reflects on her transition from private aviation to private markets and how family office expectations have evolved, from digital onboarding as innovation, to interactive live data and portfolio-level analytics as the new baseline.
[00:04:14] Inside the Investment Team: AI Meets 250 Funds Per Year
With 15 investment professionals monitoring over 250 alternative funds annually, Marine explains how AI is already accelerating document review, Excel-based financial modelling, and IC memo generation, and what the shift from efficiency to value creation will look like.
[00:05:50] When AI Became a Commercial Priority at Moonfare
Marine traces the timeline: from deploying basic AI tools nine to twelve months ago to improve team workflows, to now fielding client expectations that AI will enhance their own user experience with more interactive, live data.
[00:07:20] Lead Triage at Scale: 10,000–15,000 Registrations, 23 Markets
Marine explains how AI-powered lead scoring and triage is replacing the archaic five-to-ten-person junior sales teams that previously managed lead qualification, enabling Moonfare to focus investor relations talent on the most mature, high-value opportunities.
[00:10:00] The High-Net-Worth Line: Where AI Stops and Humans Begin
Marine draws a sharp distinction between using AI in the back office for efficiency and using it in direct client interactions. Family offices and HNW investors don’t want to talk to a bot and the human relationship remains the firm’s commercial edge.
[00:11:00] Building an AI Culture: Working Groups, Champions, and Brand Training
Marine describes Moonfare’s internal AI rollout: cross-functional working groups, team champions, and a practical project using AI to enforce brand tone-of-voice consistency across every client touchpoint.
[00:11:57] Human-in-the-Loop Is Non-Negotiable
Marine is direct: AI is a formidable shortcut, but you cannot rely on the output without human review. The investment and commercial teams treat AI as a tool to go faster, not a replacement for judgment.
[00:14:15] Compliance at 100%: AI Listening to Every Call
Previously, compliance teams could only sample a fraction of recorded calls. Marine highlights how AI now enables 100% call coverage, a step-change in risk management that was simply not achievable with human resource alone.
[00:15:20] The 5–10 Year Vision: AI Surfacing Unique Managers and Live Portfolio Tracking
Marine paints a picture of an AI-enabled investment office that can surface niche funds, track every underlying asset in live time, and eliminate the lag of quarterly reporting, replacing tribal knowledge with something close to perfection.
[00:18:00] Advice for CCOs and CIOs Who Haven’t Yet Started
Marine’s message is unambiguous: if you haven’t started, you’re behind. Get curious, register for alerts on LLM prompting, be precise about the ROI you’re looking for, and don’t just buy the most talked-about tool.
Episode Resources:
Marine Eugene on LinkedIn
Moonfare Website
Victoria Sienczewski on LinkedIn
AuumAI Website
If you enjoyed this conversation, make sure to subscribe, rate, and review on Apple Podcasts, Spotify, and YouTube.
Harry Seplowitz of Aksia explains how the $360B alternatives specialist has embedded AI across its investment process, from initial manager screening to LPA benchmarking, without losing sight of what makes its research truly differentiated. This conversation is a candid account of what AI adoption looks like inside a firm where proprietary data, and the experienced team who know how to act on it, is a real edge.
Aksia manages relationships with over 100 clients, monitors 14,000-plus distinct holdings, and reviews hundreds of investment opportunities annually. In this episode, Harry Seplowitz, Managing Director on the Pan-Alternatives team, walks through how Aksia has systematically embedded AI across investment screening, co-investment underwriting, cross-team knowledge sharing, middle office operations, and LPA review.
Harry describes how Aksia's proprietary database of 150,000+ private markets transactions spanning more than 20 years has become the fuel for AI-driven analysis. With that foundation in place, the firm can rapidly benchmark any new manager against its entire universe, identify headline risks in a manager's history before committing to full diligence, and surface the full breadth of cross-team intelligence.
But perhaps the most important takeaway is what Aksia is choosing not to do with AI. Harry makes a sharp distinction between using AI to generate longer diligence reports and using it to produce shorter, sharper, more data-driven ones. In his view, firms that let AI inflate their output are eroding, not enhancing, their value proposition.
What You'll Learn:
About the Guest:
Harry Seplowitz is a Managing Director on the Pan-Alternatives team at Aksia, based in London. He focuses on implementing customised portfolio solutions for Aksia's European and Middle Eastern clients across private equity, private credit, real assets, and hedge funds. Aksia is a leading alternative investment specialist with over $360 billion in assets under supervision and serves as a partner to many of the world's largest and most sophisticated allocators.
Episode Highlights:
[00:01:51] When AI Became Strategic at Aksia
Harry traces how Aksia's early adoption of public AI tools, driven by a tech-savvy team, quickly became a firm-wide priority.
[00:02:49] Deal Vault: Turning 150,000+ Deals into a Screening Engine
Harry describes how Aksia uses its proprietary private markets database to map every deal a manager has ever done, and benchmark it against the full universe in seconds. The result: faster conviction and faster 'no.'
[00:04:19] Co-Investments: AI as a Speed and Precision Advantage
On co-invest opportunities, Aksia can immediately surface every deal a manager has done in a specific sector and geography and compare it against all relevant deals in their proprietary database.
[00:05:16] Dissolving Team Silos: One Company Name, Full Firm Intelligence
Harry explains how typing a single ticker or company name now surfaces the complete Aksia view across hedge funds, private equity, and private credit, including every manager's long or short position, thesis, and historical context.
[00:07:42] LPA Benchmarking at Scale
With tens of thousands of LPAs reviewed across the firm's history, AI has shifted the LPA review process from summarisation to benchmarking - identifying what's off-market, what's non-standard, and how terms have evolved over time.
[00:09:45] The Middle and Back Office Dividend
Aksia's clients invest across 14,000 holdings, generating tens of thousands of annual transactions. AI-driven data extraction from capital account statements, valuations, and investor calls is cutting input errors and freeing senior time for higher-value work.
[00:10:28] Reordering the Diligence Process
AI has enabled Aksia to front-load background and company checks, identifying potential headline risks before committing weeks of cross-team resources.
[00:11:58] Measuring ROI: Report Length as a KPI
Harry describes Aksia's unconventional AI metric: they want diligence reports to get shorter, not longer. Firms letting AI inflate report length are substituting volume for insight.
[00:14:14] How Aksia Organised Its AI Rollout
Centralised for vendor selection, compliance, and data privacy. Distributed for everything else. Team champions across asset classes drive use case development and AI adoption is now embedded in end-of-year performance reviews.
[00:15:47] Junior Staff as Innovation Leaders
Harry pushes back on the narrative that AI threatens junior hiring. At Aksia, junior and mid-level employees have been the most effective AI champions, and the firm is leaning into that, giving them ownership of a strategically critical initiative early in their careers.
[00:17:32] What AI Looks Like at Aksia in 3–5 Years
The vision: producing initial insights on every institutional fund in the market at scale, driven by proprietary data. Full diligence effort reserved for what actually requires human judgment. The edge increasingly comes from pairing proprietary data with experienced human judgment to act on it.
[00:19:39] Advice for Allocators Just Getting Started
Don't start with a blank sheet. Start with the recurring task you did last week and ask how AI could have made it faster. Then ask the AI itself, it will usually suggest better applications than you would.
Episode Resources:
Harry Seplowitz on LinkedIn
Aksia Website
Victoria Sienczewski on LinkedIn
AuumAI Website
If you enjoyed this conversation, make sure to subscribe, rate, and review on Apple Podcasts, Spotify, and YouTube.
Disclaimer: This podcast is for informational purposes only. The views expressed are those of the speakers as of the recording date and may change over time.
Breanna Genecov and Kunal Koppula of Makena Capital Management discuss how the firm is rebuilding its data architecture as a prerequisite to AI integration. They detail the transition from fragmented, Excel-dependent workflows to a unified, cloud-based data stack, and explain why institutional investors who skip this foundation will struggle to extract reliable value from AI. Practical guidance on change management, leadership buy-in, and phased execution makes this essential listening for LPs & allocators at the early stages of their own transformation.
Makena Capital Management is nearly two decades into its institutional investing history and one year into a deliberate, firmwide data transformation. In this episode, Breanna Genecov (Portfolio Solutions & OCIO) and Kunal Koppula (Data Engineering) explain the technical shift from managing "Keyman risk" in isolated Excel files to building a unified cloud infrastructure. They discuss the practicalities of porting tools into a modern data stack, the necessity of firm-wide upskilling, and the phased roadmap from structured data consolidation toward AI-enabled unstructured data analysis.
The conversation cuts through AI noise: before any meaningful automation or intelligence layer can be built, your organizational data foundation must be sound. The guests detail how siloed Excel workflows, manual data pulls, and keyman risk have constrained the firm's analytical capacity, and how a centralized data stack is changing that.
What You'll Learn:
- Why Makena treats data infrastructure as the non-negotiable prerequisite for AI adoption
- How a $22B OCIO moved from fragmented, Excel-based workflows to a centralized, cloud-native data architecture
- What "one source of truth" means operationally and why inconsistent data across teams leads to incongruous decision-making
- How to manage the cultural and skills gap when moving analysts from Excel to SQL and BI tooling, without making data engineering their day job
- The two-phased roadmap: structured data consolidation first, then unstructured data contextualization via AI
- Why "garbage in, garbage out" is the most important AI principle institutional LPs & alloactors aren't taking seriously enough
- What leadership buy-in actually looks like in practice and why without it, transformation stalls
If you enjoyed this conversation make sure to subscribe, rate, and review it on Apple Podcasts, Spotify, and YouTube.
About the guests:
Breanna Genecov leads the Portfolio Solutions and OCIO team at Makena Capital Management. Over more than a decade at the firm, she has held roles across manager research, portfolio strategy, and risk management, with a current focus on the design and implementation of multi-asset class portfolios.
Kunal Koppula leads Makena's data engineering team. He previously worked as a trader and quantitative developer before joining Makena, where he now oversees all data-related workflows across the organization and is spearheading the firm's transition to a centralized, scalable data architecture.
Episode highlights:
[00:02:05] The Catalyst for Change
Breanna traces Makena's data evolution from a 2016 platform overhaul with uneven adoption, to the current firmwide initiative driven by leadership.
[00:03:16] What the Data Transformation Entails
Kunal outlines the scope: consolidating all data into a single source, replacing keyman-dependent Excel processes with automated cloud workflows, and building a unified data lineage across every team.
[00:05:29] The Manual Tax Described in Detail
The guests describe the true cost of legacy workflows: manual PDF and portal extraction, repeated Excel downloads, and the compounding risk of human error across every reporting cycle.
[00:08:44] Breaking Out of Silos Across Investment, Operations, and Client Teams
Kunal explains how his role as “connective tissue” across teams surfaces redundant workflows and enforces a single source of truth.
[00:14:53] Phase One vs. Phase Two: Structured and Unstructured Data
Kunal distinguishes between the current phase of structured data consolidation, and the emerging phase: making unstructured data (PDFs, manager notes, etc.) accessible and contextualizable via AI.
[00:17:20] Advice for Peers Jumping Straight to AI
Breanna and Kunal make the case against skipping the foundation which they say includes covering garbage-in/garbage-out risk, the need for human validation, the importance of the right project lead, and why patience is a strategic requirement.
Episode resources:
- Breanna Genecov on LinkedIn
- Kunal Koppula on LinkedIn
- Makena Capital Management Website
- Victoria Sienczewski on LinkedIn
- AuumAI Website
Chris Hartnoll, CEO and Managing Director of HICO Investment Group, discusses how the firm is approaching AI adoption across its multi-asset class endowment style portfolio and direct investment platform. He explains why preparation, security, and operational groundwork must come before any meaningful AI deployment, and shares early signs of productivity gains and improved risk identification. Practical guidance on experimentation, leadership mindset, and the urgency of starting now makes this essential listening for LPs and allocators considering their own AI journey.
Chris Hartnoll operates at the intersection of institutional investing and industrial technology. At HICO Investment Group, he has built a global multi-billion dollar platform that pairs a sophisticated multi-asset class endowment style portfolio with direct investment and incubation of new ventures in sectors like maritime logistics and energy transition. In this episode, Chris unpacks what it means to prepare an investment office for AI: the security considerations, the folder structure decisions, the legal questions around data protection, and the cultural shift required to move from manual processes to AI-assisted workflows.
The conversation is grounded in Chris’s background navigating high-stakes environments, from aerospace engineering at Rolls-Royce to Morgan Stanley M&A Advisory to the Royal Marines. He draws a direct line between the discipline of preparation in those fields and the approach HICO has taken to AI adoption: do the unglamorous operational work first, then experiment in discrete, measurable process steps. He argues that the allocator community has been slow to look inward and apply AI to its own processes, despite readily advising GPs and portfolio companies to do the same.
What You’ll Learn:
If you enjoyed this conversation make sure to subscribe, rate, and review it on Apple Podcasts, Spotify, and YouTube.
About the guest:
Chris Hartnoll is the CEO and Managing Director of HICO Investment Group, a global investment firm focused on maritime, logistics, and energy transition. Chris built and oversees a multi-billion dollar platform that includes a multi-asset class endowment style portfolio paired with direct investment and incubation of new ventures. Before founding HICO, Chris held roles in aerospace engineering at Rolls-Royce, M&A advisory at Morgan Stanley, and served as a Royal Marine. He holds an MBA from Harvard Business School.
Episode highlights:
[00:01:46] Background and Approach to Risk
Chris traces the thread connecting his Royal Marines experience, aerospace engineering at Rolls-Royce, and Morgan Stanley M&A to how he thinks about preparation and risk management in investing.
[00:03:31] Balancing Endowment Style Asset Allocation with Direct Investment
Chris explains how HICO balances its capital allocation process and its direct deal-making and incubation activities, and why keeping the processes separate while sharing information has been key.
[00:05:17] Why the Allocator Community Has Been Slow to Adopt AI
Chris reflects on why investment professionals are quick to advise others on improvement but slow to apply the same lens to their own processes, and why CIOs have been reluctant to disrupt what they’ve built.
[00:07:27] The Operational Groundwork for AI Adoption
Chris details the unglamorous but critical preparatory work: security guardrails, data privacy, folder structure, legal review of GP document protections, and building the right digital infrastructure.
[00:08:55] Early Signs of Value: Productivity and Risk Identification
Chris shares where AI is already delivering: summarizing LPA documents, capturing manager interview data, and surfacing new ways of viewing risk.
[00:12:57] The Future of the AI-Enabled Investment Office
Chris outlines his vision for human-AI collaboration in investing, why private markets will continue to have a human-in-the-loop, and how AI-enabled teams will gain a decisive speed advantage during periods of market volatility.
[00:15:39] Advice for Peers on the Fence
Chris delivers his core message: be prepared first, then start experimenting immediately, because the learning curve is long and the cost of waiting is compounding.
Episode resources:
Today, we are introducing The Agentic Allocator, an exclusive conversation series that pulls back the curtain on how the world's most forward-thinking LPs & allocators are actually deploying AI in their investment offices.
For many LPs & allocators, implementing AI still feels like a "black box". In this series we dig into what it really looks like to move beyond legacy workflows toward agentic AI, and the real obstacles allocators are navigating along the way.
If you’re assessing how AI might look like inside your investment office and where to start, you can find more resources here.
From the publisher's feed