The PE Recruiting Machine Just Broke. Nobody's Talking About the Real Casualty.

Apollo paused 2027 associate hiring. General Atlantic followed the next day. TPG, then KKR fell in line. Goldman Sachs now requires junior analysts to sign quarterly loyalty attestations confirming they have not accepted an outside offer. JPMorgan threatened termination for anyone who arrived with a PE offer already in hand.
The industry framed this as a talent war. Compensation arms race. Structural reform of the on-cycle recruiting timeline.
All accurate. All missing the point.
The real story is not about who gets hired. It is about what gets lost every time someone leaves.

The average PE analyst tenure before moving to a new fund, launching a company, or going back for an MBA sits somewhere between 18 and 36 months. That window is exactly long enough to learn the firm's sector coverage, internalize the investment thesis, build a read on a dozen industries, and run two or three deals end to end.
Then they leave. And the next person starts from scratch.
This is not a complaint about turnover. It is an observation about architecture. The knowledge those analysts built does not live in the CRM. It does not live in the model. It lives in the thousands of documents they read, the patterns they noticed across deals, the red flags they mentally catalogued but never formally wrote down. That context is invisible infrastructure. And right now, every time an analyst walks out the door, that infrastructure walks out with them.
85% of analysts leave investment banks within their first two years. PE firms are not immune. The recruiting chaos of the past 18 months has made the cycle faster, not slower.

The standard response to this problem is "better documentation." It never works. Nobody writes down their mental models in real time. Nobody creates a structured record of why they passed on a deal at the screening stage. The friction is too high, the incentive is too low, and the output would be unreadable anyway.
The better question is: what if the documents themselves became the memory?
Every deal a team looks at leaves a trail. CIMs, management presentations, data room exports, IC memos, market research. That material contains everything an incoming analyst needs to understand why a deal was interesting, what the concerns were, what comparable situations looked like. The problem is it sits in a folder nobody searches, in a format nobody can interrogate at scale.
This is the actual unsolved problem in knowledge work at finance firms. Not that people leave. That when they leave, the research disappears.

The firms building a structural answer to this are mostly enterprise-scale. Hebbia has 33% of the top global asset managers as clients. The tooling is serious. The price tag matches: we are talking Bloomberg Terminal territory, custom implementation, months-long sales cycles, data stored in the US.
For a mid-market PE fund in Lyon, a credit team in Amsterdam, or an M&A boutique in Vienna, that is not a realistic option.
Which is part of why the problem persists. The tools that exist are built for the top of the market. The rest of the industry keeps losing institutional knowledge one analyst rotation at a time.

I am building Lens to close that gap. Document intelligence for European mid-market finance teams. Self-serve, GDPR-native, priced for teams rather than enterprises.
The core idea is that your deal library should outlast your headcount. When the analyst who ran your last 8 CIM reviews leaves, their work stays searchable. The next person can ask what was looked at in the healthcare sector in 2023, what the common red flags were, what the pass rationale said, and get cited answers pointing to the exact source.
The recruiting machine breaking is an industry problem. The knowledge architecture staying broken is a choice.

Early access is open. If you run a finance team in Europe and this maps to something real in your workflow, I want to talk.