The Art of the Possible: What a Room Full of Senior Fund Principals and Lawyers Told Us About Structural Complexity

Last month, StructureFlow brought together senior fund principals and the lawyers who advise them for an evening of candid, Chatham House–basis conversation in New York, hosted with Andrew Forman of TD Securities. The premise was simple: put people from both sides of the table in the same room and ask them to say what actually is and isn’t working. 

What came back wasn’t a tidy list of talking points. It was four threads that kept resurfacing all night, each sharper than we expected. 

Customization Is the Real Story, Not Complexity for Its Own Sake 

The tax that no one in the room disputed is that fund structures are getting more complex. But the more interesting debate wasn’t whether complexity is rising, it was what’s actually driving it. The room’s answer: customization. A hard fundraising market has pushed limited partners to ask for more—fund-of-one structures, side letters, separate accounts, bespoke terms–and sponsors have grown more willing to say yes. Add evergreen and semi-liquid structures, plus the financing now layered onto nearly every limited partner interest, and a structure gets more intricate simply because the deal itself got more specific. 

One investment funds partner put it plainly: clients aren’t asking for complexity, they’re asking for customization, and a little more of it accepted early often heads off far more of it later. 

The market isn’t converging on simpler structures anytime soon. The opportunity is to make bespoke structuring more efficient to manage at scale. 

AI Has Reset the Clock Faster Than Judgment Can Follow 

This is where the conversation got animated. Several participants described deal timelines compressing from weeks to days. One recalled a transaction moving from initial inquiry to signed closing within a week, with documents reviewed in hours rather than days. Nobody in the room thought that pace made the work better. Several thought it simply moved the risk into a new place: the duty of care to validate what AI surfaces before anyone acts on it. Many noted that this is creating a new cost tension: if AI halves the time, firms cannot charge twice as much. 

A participant who leads applications and AI strategy for a major firm’s private capital practice offered the sharpest framing of the night. Judgment can’t be simulated, because negotiation runs on things no dataset captures. AI multiplies whatever judgment, context, and infrastructure you already bring to the table, and little else. The same participant described handing an AI-generated deal analysis to an experienced lawyer and watching it land as entirely correct and entirely beside the point: accurate answers to questions nobody needed asked, because whoever prompted it didn’t know what to ask. 

This compression isn’t likely to level off. The more useful response is freeing up time for the judgment-intensive work that can’t be sped up, not just moving faster on the parts that can. 

A Single Source of Truth Doesn’t Exist, and No One Agrees Whose Job It Is to Build One 

Almost everyone in the room agreed on something surprising: a single source of truth for entity and structural data doesn’t exist, for anyone at the table. What followed wasn’t agreement on a fix. It was a live argument about ownership. Is it the advisor’s job, since advisors interpret the structure? Or the fund’s, and if so, which function – tax, legal, compliance, risk, and reporting each keep their own version. 

One participant, who has worked on both the vendor and law firm side of entity data, called this the most stubborn unresolved problem in the room. And the gap runs both directions. A fund-side participant admitted their internal data is well organized, but it only reaches outside counsel as an occasional, static update, never as something live or interactive. Advisors described the mirror image: plenty of tools exist to interpret a structure after a deal closes, and almost nothing helps plan one beforehand. 

Another participant raised a version of this that landed close to home for the room: what happens when a fund’s structure has to be reconstructed years after formation, once the people who built it have moved on and taken the institutional knowledge with them. 

Where this likely heads is way from the periodic handoff or reconstruction years later toward a live model both sides can work from as the structure evolves. That shift, from reconstructing understanding to sharing it in real time, is what closes this gap. 

Judgment Is the One Thing Nobody in the Room Believed AI Could Replace 

There was no dissent on this point, and no hedging either. Lawyers spend their day answering two questions: what should I do, and how do I do it. AI can help with the second. It has no claim on the first. You can’t simulate the other side of a negotiation, and every lawyer in the room had lived some version of the moment where “I understand your point, it makes total sense” gets followed by “my client says it’s a deal breaker.” 

AI is a force multiplier for people who already bring judgment, context, and infrastructure to the table. For everyone else, it produces noise dressed up as an answer. 

The tools that hold up here will be the ones built to augment human judgment, not replace it. That distinction is likely to become the line between what gets adopted and what gets shelved. 

Where This Leaves Fund Managers and Their Lawyers 

Put the four threads together and a clear picture forms. Structural complexity isn’t reversing, and nobody in the room expects it to. What can change is the tolerance for reconstructing an understanding of that structure every time it matters: across a negotiation, a financing, or a decade of amendments nobody wrote down in one place. 

We didn’t walk in with a thesis to prove, and we didn’t walk out with one either. We walked out with better questions. Who should own structural and entity data? Where does healthy customization end and unnecessary complexity begin? What would it actually take for a fund and its advisors to work from one live model instead of a periodic handoff? 

If any of this sounds familiar, we’d like to hear from you. 

Stay Connected 

This is one conversation in an ongoing series with fund principals and their advisors on where structural complexity is headed. We’ll be sharing more of what we’re hearing, always without names or firms attached. If you’d like to be part of the next one, reach out to your StructureFlow contact or schedule a meeting.