Every fund has a single point of failure.

Six years ago, a fund's VP of investor relations took over coverage of its largest limited partners, including Harrison Capital, the fund's biggest commitment. She onboarded new investors, negotiated their side letters, sat through their quarterly calls, and became the person each of them called first when something on a capital call notice didn't look right. Late last spring, she resigned. In the two weeks before her last day, the firm tried to capture what she knew — which investors had informal fee arrangements, which ones expected a phone call before a wire and which were fine with an email, which relationships carried enough history that a routine request read very differently depending on who was asking. Almost none of it made it into writing, not because anyone was careless, but because six years of judgment doesn't compress into a handover document written in ten business days.

What the firm was left with looked complete on paper — the CRM had every commitment, the shared drive had every subscription agreement and side letter, the email archive held years of correspondence with each investor — and on any measure a compliance checklist would use, nothing was missing. None of it, though, carried the judgment that connected those records to the moment that mattered: whether a fee offset had been promised or only floated in conversation, which investor's terse one-line emails were just their normal register and which meant something was wrong, which questions had already been asked and answered in a call, long before, that nobody wrote down.

Why the knowledge concentrates in one person

A person becomes the fastest, cheapest place to store context in the moment it's created, simply because writing something down takes longer than remembering it. A side conversation after a quarterly call, a verbal exception granted to smooth over a late capital call, a read on how a specific LP prefers to be communicated with — none of it looks, at the time, like something worth documenting. It only becomes valuable in hindsight, once someone needs the answer and the person who has it is no longer in the room. That's how a relationship ends up depending on one person's memory: through hundreds of small moments where remembering was simply cheaper than writing it down, repeated over years until the dependency was total.

The concentration is invisible for most of a relationship's life. A fund with one person covering an investor for years doesn't feel exposed; it feels efficient — that person can answer any question about the relationship faster than a system could, because they don't need to look anything up. The cost only becomes visible on the day the person leaves, and by then there's no cheap way to spread it. Knowledge that could have been distributed across six years instead has to be reconstructed inside a transition window measured in days, out of whatever fragments happen to survive in writing.

What a replacement can't inherit

Replacing the person doesn't solve this. A strong hire can absorb the CRM, read the file, and learn the fund's process within a few weeks. What they can't do is retroactively acquire years of context about a specific relationship — which LP's terse emails are just how they write, and which signal that something is wrong. The gap usually surfaces on the first call the new coverage person makes to a long-tenured investor, and everyone on the line can tell.

Brightfield Family Office has been asking a version of the same question for three years running. For three years, someone remembered the answer without having to ask what the question meant.

How fast the context disappears

The knowledge that survives a departure doesn't disappear all at once — it decays. In the first few weeks, most of it is still technically recoverable: the departed employee will usually still take a call, teammates who sat in on some of the same meetings remember fragments, and the email trail is recent enough to search with some confidence about what to look for. A few months later, the same information is much harder to reach. The departed employee has moved on to a new job and stopped fielding questions about a fund they no longer work for, the colleagues who remembered fragments have half-forgotten them or left themselves, and the emails are technically still there but scattered across hundreds of threads nobody remembers how to search.

This is a knowable risk, not an unpredictable one. A fund can identify, today, which relationships are covered by exactly one person, and how long that person has been the only one who really knows the history. Almost no firm has asked the question, because nothing forces the issue until someone hands in their notice. By then, the honest answer to what the fund knows about that relationship is whatever one person can recall before their last day.

It's worth asking how much of a fund's institutional memory depends on one person right now, and how anyone would find out the answer without waiting for that person to resign.