Ask a fund controller when they last knew, with certainty, that their internal numbers matched what their administrator had on file, and the honest answer is usually "the last time the administrator sent a trial balance." Everything in between — the trades, the accruals, the fees, the capital calls that hit the ledger every single day — runs on faith that it'll tie out when the numbers finally arrive. Most of the time it does. The times it doesn't are the times that turn into a bad afternoon three weeks after the period already closed.
This isn't a knock on administrators. A good administrator is exactly what most funds should keep as the official book — experienced hands, an independent set of eyes, a name LPs recognize on the audit. The gap isn't in who's doing the accounting. It's in the time between deliveries, where a manager has no way to independently confirm anything.
The reconciliation everyone does once, by hand, under deadline
When the administrator's trial balance does land, reconciling it against the fund's own records is still, for most managers, a spreadsheet exercise: paste one column next to another, eyeball the differences, chase down whichever ones are large enough to notice. It's tedious, it's manual, and it only happens on the administrator's schedule — which means a real break, an accrual booked in the wrong period, an expense that never made it into one side of the ledger, sits there unnoticed for as long as the delivery cycle allows.
The fix isn't to reconcile faster by hand. It's to stop waiting for the delivery to start checking.
What we built
FundOS now computes its own trial balance from the fund's live general ledger every day, and reconciles it — account by account — against whatever trial balance the administrator delivers, on whatever cadence they deliver it. The administrator stays the official book. FundOS becomes the independent, always-current second opinion that either confirms it or catches the specific account where the two disagree.
The part that used to take a person the longest — mapping the administrator's chart of accounts onto the fund's own — happens once. The first time a fund uploads an administrator statement, the system proposes a mapping: which administrator account corresponds to which internal account. A human confirms it. From that point forward, every future delivery reconciles automatically against that confirmed mapping, and any account the administrator introduces that wasn't there before gets flagged rather than silently guessed at.
The trust doesn't come from an AI deciding the books agree. It comes from a human confirming the mapping once, and deterministic code checking the arithmetic every day after that.
Every break gets a reason, not just a red flag
A number that doesn't match is not, by itself, useful information. What a controller actually needs to know is why it doesn't match, because the answer changes what happens next. An accrual that's timing-related will true itself up next period and doesn't need a phone call. An expense missing from one side of the ledger needs one today. A currency translation difference needs a different kind of look than an account nobody mapped yet.
So every break gets classified — accrual timing, a missing expense, an FX translation difference, or a mapping gap that still needs a decision — and lands in the same exceptions inbox that already surfaces covenant breaches, settlement fails, and filing deadlines. Nothing gets buried in a spreadsheet tab. And because the reconciliation runs daily, a break that resolves itself — the accrual that catches up, the account that gets remapped — clears on its own the next time the numbers agree, instead of sitting open until someone remembers to close it out.
Nobody has to rip anything out to start
The single biggest reason funds don't adopt better back-office tooling is that adopting it usually means a migration — cut over the books, hope nothing breaks, live with the risk for a quarter. Shadow reconciliation doesn't ask for that. A fund uploads its administrator's most recent signed trial balance, the system builds an opening position from it, and FundOS starts shadowing from that exact day forward. Nothing about the fund's relationship with its administrator changes. The only new thing is that, every day, someone — or something — is checking the two books against each other instead of hoping they agree.
For a fund that eventually decides it wants FundOS to be the book of record rather than the shadow of one, the path is the same tooling, one setting away — no separate migration project, no second implementation. But that's a decision a fund makes on its own timeline, with months of matching evidence behind it, not a sales pitch it has to take on faith.
The number your administrator sends you is very likely correct. The point was never to doubt it. It's that a fund shouldn't have to wait until the email arrives to find out.