Every organization carries a body of knowledge that exists nowhere in writing. That is known as institutional knowledge, and family offices are no different. Why a particular legal structure was chosen. How a counterparty relationship evolved over a decade. The unwritten preferences that shape what the family will and won’t consider. The reasoning behind a decision made three years ago that still governs strategy today. This knowledge sits in people’s heads, in email threads, in conversations that were never documented. In most offices, it works. The teams are small. The people tend to stay. And the knowledge moves through the office the way it always has: through experience, repetition, and proximity.
Interested in learning more about what’s actually possible with AI and the options for family offices?
But the risk is equally familiar. In a team of five or six people, two or three individuals typically hold the vast majority of institutional knowledge. They know why things are done a certain way. They carry the context that makes decisions coherent. When one of them leaves, retires, or is unavailable for an extended period, the knowledge goes with them. A Dentons survey of over 200 family offices found that 54% identified key-person turnover as their top internal threat, ahead of cybersecurity, talent retention, and privacy. This is key person risk in its most fundamental form, not a skills gap, but a memory gap.
A problem the office has never quite solved
Family offices have talked about knowledge management for years. The solutions have always fallen short: shared drives that nobody maintains, internal wikis that go stale within months, onboarding documents that capture process but not judgment. EY has observed that many offices fall into what it calls “process by trust,” where long-serving professionals are relied on for their institutional knowledge rather than backstopped through documented processes. Documentation may exist for legal agreements and service contracts, but rarely for the core operational knowledge that keeps the office running. The difficulty is that institutional memory isn’t really about documents. It’s about context, reasoning, relationships, and accumulated understanding. The people who hold this knowledge rarely think of it as knowledge. It’s just how they work. And because it lives in their heads, it has never been systematically captured. Not because the office didn’t want to. Because there was never a practical way to do it.
Something is changing
AI is becoming the first tool that can genuinely hold this kind of knowledge. Not in theory, but in practice, through normal daily use.
When a team member works with an AI assistant over weeks and months, something new is happening. The analyst who has spent six months working with an AI tool on portfolio analysis has built a body of accumulated reasoning that the office has never had before: documented, searchable, reusable. The operations lead whose conversation history contains the logic behind compliance decisions has, without intending to, created a record of institutional judgment that would otherwise exist only in their head. The person drafting family communications through an AI tool has captured tone, preferences, and relationship context that no onboarding document could replicate.
This is institutional memory being externalised for the first time. Not through a formal knowledge management initiative. Through the simple act of working with a tool that remembers.
The key person risk starts to shift
This matters most where the risk has always been highest. In a small office, losing the person who holds critical knowledge has always meant losing the knowledge itself. Succession planning, when it exists, tends to focus on roles and responsibilities. The knowledge those people carry, the context that makes them effective, has never been part of the handover because it was never written down.
AI changes that equation. If the office’s institutional knowledge is accumulating inside AI tools as part of daily work, then for the first time, some of that knowledge exists independently of the person who created it. The reasoning is captured. The context is retained. The patterns are documented. Not perfectly, and not completely, but meaningfully. And if that knowledge lives on infrastructure the office controls, it stays with the office rather than with the provider.
For a family office, this is a significant shift. The most valuable and most vulnerable knowledge the office holds, the kind that lives in two or three people’s heads, is starting to find a home in systems that can retain it.
A foundation worth recognising
None of this happens automatically. The quality of what gets captured depends on how the tools are used, how much context is shared, and whether the office treats this as something worth being intentional about. But the starting point is already here. Teams are building institutional memory through their AI tools every day, often without realising it.
The offices that recognise what’s happening have an opportunity: to treat AI not just as a productivity tool, but as the foundation for something the office has always needed and never managed to build. A living institutional memory that doesn’t depend entirely on any single person being in the room. The knowledge has always been there. It just finally has somewhere to live.