Inside the private credit collapse, the locked funds, and the AI that saw it coming — before the fund manager did.
Casper Zhao didn't write this for general readers. He wrote it for the five people sitting on different sides of the same broken information chain.
You submitted a redemption request in the last 24 months and it wasn't fully honored. The fund manager cited the gate provision. You read the section. You understood it. You weren't satisfied.
You've granted your lender access to your reporting — and now your CRM — in ways that would've been inconceivable when the deal closed. You're managing a relationship that shifted from partnership to surveillance.
You know some of your 2021 vintage exits aren't happening the way the model assumed. You've been writing careful LP language for two years. You're starting to wonder what happens when the careful language runs out.
Your LP clients have questions about fund valuations you've been answering manually. You've wondered whether AI could do it faster and more rigorously. This book is the answer: yes — and it's a two-hour job, not a six-week one.
Not your fund specifically. The category your fund belongs to. If you've read that phrase more than twice, this book was written about the situation you're already inside of — you just haven't seen the numbers yet.
This isn't a textbook with a thesis bolted on. It's built from composite scenes inside the rooms where the decisions actually get made.
A Chicago LP meeting where a fractional CFO is quietly reading a report on her laptop that the fund manager doesn't know exists — and the entire book unfolds from what's on her screen.
The PE fund manager, the lender, the portfolio company CFO, the LP, the restructuring advisor, and the fractional CFO each tell their version. None of them are lying. All of them are withholding.
Not theory — the exact language used to make an AI tool find a $26 million valuation gap inside a quarterly report that three years of analysts had already read and missed.
ASC 470-50, troubled debt restructuring, NAV lending mechanics — the technical content is accurate and citable. The characters are composites. You get the real mechanics without the legal risk of a single identifiable target.
Every decision in this book is rational. That's the argument: a system can produce a $300 billion problem without a single person doing anything that looks, from the inside, like wrongdoing.
From the LP's first redemption request to the eleven-page legal explanation of why it wasn't honored — watch exactly how the fine print you didn't read becomes the wall you can't get past.
A four-hour restructuring negotiation, told from inside the room — where the people deciding your fund's fate don't all have the same information, and some of them know exactly what they're not saying.
Real interest coverage ratios, real exit multiples, real PIK accrual math — worked line by line so you can run the same numbers on your own fund before you finish the chapter.
Six and eighteen months later, the book checks back in — what actually happened to the CFO, the foundation, the index — instead of leaving the story conveniently unresolved.
Five things to do this week if you're an LP, a CFO, or a finance professional inside the industry — plus the actual document request list and AI prompts to run the analysis yourself.
This isn't a hypothetical number. It's built from publicly available BDC quarterly filings, LP advisory board disclosures, and private credit industry valuation surveys — the same public data Lena Park's AI stack used to find the gap on a single fund in two hours.
The uncertainty band is wide. The direction is not uncertain.
Gather what you're entitled to. Build a position-level model. Compare it against BDC filings and secondary pricing. Characterize the gap as a question, not an accusation. This is the same method described across every chapter, broken down by role in the appendix.
The specific instructions used to surface a management-fee timing pattern and a 28% valuation discrepancy — written for general-purpose AI tools, not a proprietary platform you have to buy.
Direct lending, mezzanine, NAV lending, gate provisions, in-kind redemption, TDR, Level 3 assets — Appendix A turns the fine print into a working vocabulary in under twenty pages.
Redemption mechanics, PIK and distribution language, NAV lending authority, valuation committee composition, clawback terms, side letter MFN clauses — mapped to the page, not buried in a 340-page document.
Non-accrual rates, PIK income ratios, interest coverage — the four data fields that let you benchmark your "performing" private credit fund against publicly audited comparables in an afternoon.
Is PIK income over 15% of total investment income? Have distributions exceeded cash interest income? Has any position had a covenant amendment in the last 12 months? Three yes/no questions that tell you whether to commission the longer analysis.
The single accounting judgment call — under ASC 470-50 — that decides whether your fund recognizes a loss or carries it forward indefinitely, walked through from inside the credit committee that makes it.
$25,000–$75,000 for a full review on a $40M position — a number you can size against your own allocation before you decide whether to commission one.
Separate, specific checklists for the LP, the portfolio company CFO, and the finance professional inside the industry. No generic advice — every action ties back to a chapter that shows you why it matters.
The book's real claim isn't that private credit is collapsing. It's that the cost of verifying a mark has fallen and isn't going back up — which means the choice to find out is now yours, not the GP's.
[ tap a redacted line to declassify it ]
Every decision was rational. The aggregate was not. That is what a credit cycle looks like from the inside.
You were not paid for patience. You were paid for not asking.
The standard has a correct answer. The room has a preferred answer. They are rarely the same.
The third waiver is when you realize this is no longer a financing problem. It is a business problem dressed in financing language.
The GP-led secondary is how you extend the problem while telling LPs you solved it.
The model was not wrong. The assumptions were just the ones the GP needed.
The covenant was the protection. The competition to win the deal was the reason it was not there.
The LP agreement was written to be unread. When it was finally read, it worked exactly as designed.
She did not find fraud. She found assumptions. That was enough.
A zombie portfolio survives by being expensive to verify. That was never a permanent condition; it was a cost structure, and cost structures change.
Whether you're holding the capital, managing the fund, or sitting in the CFO's chair between them — this is the book that tells you what your numbers are actually saying.