December, 2024
As private investments grow in complexity and popularity, family offices and institutions are confronting a new set of operational and strategic challenges. At the November 2024 FOTechHub virtual conference, four industry leaders came together to unpack how technology is helping (and sometimes hindering) the private investment ecosystem. The discussion centered on the evolving role of artificial intelligence (AI), automation, and digital infrastructure in managing alternatives—and what family offices should do next.
The panel included Danielle Roseman, Co-Head of Lazard Family Office Services; John Jennings, President and Chief Strategist at St. Louis Trust & Family Office; Joe Larizza, Managing Director of Data Solutions at iCapital; and Paul Williams, SVP and Product Lead for Alternative Fund Services at Northern Trust. Each brought practical insight into how their firms are using—or resisting—AI and automation to navigate the operational demands of private investing.
Efficiency, Not Replacement: What AI Can Actually Do Today
Much of the discussion explored the real vs. hyped capabilities of AI in the alternatives space. John Jennings was frank: "We're still experimenting. We're not going to put anything in front of clients that could compromise confidentiality or accuracy." While the public narrative often centers on AI replacing advisors, the panelists described a more realistic use case: internal productivity.
Joe Larizza agreed, noting that "machine learning is further ahead than large language models" when it comes to reliability. His team at iCapital uses AI to automate document processing for over 100,000 alternative investments monthly, but it still requires human oversight. "We’re at about 70% accuracy—so there's still a lot of human work in the loop."
Alternative Structures, Alternative Frictions
As investor appetite for private investments increases, the types of vehicles available have multiplied—from evergreen funds to semi-liquid BDCs. Paul Williams pointed out that while these products aim to increase access and flexibility, they bring operational headaches: "You're trying to match very different data timeliness, precision, and liquidity expectations."
GPs, he noted, are realizing that digitization isn’t just about investor engagement, but also about their own cost structures and operations. Whether through digitizing fund documents or leveraging natural language processing to draft fund commentaries, many private fund managers are prioritizing tech to gain an edge on both distribution and cost control.
Digitizing the Back Office: Low-Hanging Fruit
One consensus point? Statement processing is a clear AI win. As Danielle put it, "It used to take a Mirador or iCapital to make sense of all that reporting. Now an individual advisor or analyst can start extracting and summarizing that data faster."
Joe emphasized the irony in this space: “We’re spending billions of dollars trying to read things that were created in structured data and then got converted to PDFs. Let’s fix the source.”
The discussion also touched on use cases like Know Your Client (KYC) and Anti-Money Laundering (AML) processes, which remain stubbornly manual. AI can help, but the real gain comes from addressing the format and flow of data upstream.
Human Advisors Are Still the Filter
Despite AI's growing capabilities, the panelists were unified in their view that human advisors aren't going anywhere. As John explained, "There’s a difference between experience and expertise. If advisors aren't learning from feedback, they won’t be replaced by AI—they'll be replaced by other advisors who are."
Joe agreed, likening AI's role to that of the robo-advisor moment. "It didn’t eliminate advisors; it gave them a better toolbox. The best users of AI are still humans who know what to ask."
Public vs. Private LLMs: A Trust Issue
The panel concluded with a conversation around private vs. public AI models. While tools like Microsoft CoPilot offer hybrid approaches, confidentiality remains paramount.
Danielle shared that Lazard is experimenting with a private LLM but still uses public models like ChatGPT for broader research. "There’s a sense of safety in private models, but they can’t yet process the same volume of data. You trade confidentiality for capability."
John underscored the point: "We have 65 employees and 64 client families. We just don't have enough data to train a private model in a meaningful way."
Final Thought
AI isn’t the disruptor some feared—yet. But it’s already proving to be an enabler of speed, scale, and smarter decision-making in the private investment space. Family offices that lean into practical, back-office use cases while maintaining a healthy skepticism toward hype are best positioned to thrive.
This article is based on a panel discussion at the FOTechHub 2024 Virtual Conference.