Ask a fund CFO what their fund administrator does and you'll get a confident answer. Ask what it doesn't do and the answer gets vague — not because the CFO is unclear, but because the unowned work doesn't have a name. It's just "the rest of it," and it lives in their inbox.

This is a map of the boundary. It's useful for deciding what to buy, and more useful for deciding what not to.

What the fund administrator owns

This part is well-defined and, at a good admin, well-executed:

The books of the fund — general ledger, capital accounts, allocations
Capital calls and distributions — notices, mechanics, tracking
Quarterly and annual financial statements — the reporting package
Performance calculations — IRR, TVPI, DPI, the waterfall
The LP portal — where investors go for statements and documents
Audit support — workpapers for the auditor

If you're paying an administrator and not getting these cleanly, that's an administrator problem, not a software gap. No tool fixes it.

What the LP portal owns

Narrower than people assume. A portal is a distribution surface: it's where a document goes after it exists. It stores statements, holds tax documents, records who accessed what, and handles the notice.

What a portal does not do is produce anything. It doesn't answer a question. It doesn't reconcile. It's a well-organised filing cabinet with an access log, and it should be judged as one.

What the administrator will do, but charges for

Then there's a middle tier — real services, priced as add-ons, which is where the economics get uncomfortable at sub-$1B:

KYC/AML on LPs — commonly $100–150 per investor, which is a real number when you're onboarding a fund
Management company accounting — most admins will do it; most funds find the pricing hard to justify
Portfolio monitoring — only the newer, AI-forward administrators offer this at all

None of this is a criticism of administrators. These are genuine services with genuine cost. But for a firm managing $100M with a CFO and an ops lead, the per-unit pricing frequently doesn't clear the bar, so the work comes back in-house by default rather than by decision.

The five things nobody owns

Here's the actual gap. Every item below is recurring, time-expensive, and sits outside both the administrator's scope and the portal's:

1. Due diligence questionnaires

Ten-plus per fundraise. Around 80% the same questions in different phrasing. The worst institutional one can run one to two weeks. Your administrator doesn't answer these — they're statements about your firm, its policies, its people, its process. Only you have the source material.

2. The LP inbox

The single largest consumer of a fund CFO's week isn't close. It's responding to email — LPs and internal team members asking for documents and fund information that already exists somewhere. The portal has the document. The email still arrives, because the LP would rather ask than look.

3. K-1 review

Your administrator or tax preparer produces the K-1s. Someone still has to check them against the partner capital accounts before they go out — allocation timing, mid-year transfers, book-versus-tax differences, bundled partners. That review is yours. It arrives in a two-week window in September with a hard deadline.

4. Management company finance

Payroll, expense allocation across entities, the manco P&L, budget-versus-actual, bill.com and expense-tool categorisation into a chart of accounts. This is a real finance function, it just isn't a fund finance function, so it falls outside the administration engagement.

5. The review of the administrator's own work

This one is invisible on every org chart and it's the one CFOs describe most bitterly.

Quarterly close for a sub-$1B fund runs roughly 20 hours: send valuations, review the administrator's workpapers, review the capital statements. The variance isn't in your work — it's in theirs. A good administrator produces a package you review once. A struggling one produces multiple rounds, and each round is your time.

Nobody sells software for "checking your fund administrator." But it's a recurring job with a deadline, it requires reading two documents against each other, and it is precisely the kind of work that deterministic checks are good at and tired humans at quarter-end are not.

Why this gap exists

It's structural, not accidental.

Administrators are priced and staffed around fund-level, per-fund, recurring work with defined deliverables. Portals are priced around investor count and document volume. Both scale on axes that have nothing to do with how much questionnaire volume you have, how noisy your LP base is, or how many entities your management company runs.

So work that doesn't fit either axis lands on the person with no axis at all: the CFO. At a sub-$1B firm that's usually a team of one or two, with a managing partner who wants the task done and doesn't want to deal with it.

What to actually do with this map

If your fund-level books are the problem — statements late, numbers wrong, multiple rounds every quarter — that's an administrator conversation, possibly an administrator change. Don't buy software for it.

If your LPs can't find documents — that's a portal or a process problem. Also not software.

If your week is questionnaires, inbox, K-1 review, manco finance, and re-reviewing your administrator's package — that's the gap, and it's the only part where tooling changes the arithmetic.

Fundry is built for the gap and nothing else. We don't keep the fund's books and we don't host your LP portal — those are done better by people who do only that. What we do is the recurring work on either side of them: drafting questionnaire answers from your own documents, triaging the LP inbox, reconciling K-1s to capital accounts, running the management company's numbers, and checking the close package before you sign off on it.

Every answer is cited to the document it came from. Nothing sends without you approving it.

A note to fund administrators

If you administer funds and you've read this far: none of the above is a pitch against you. The five gap items are work your clients are doing regardless, and doing badly at 11pm.

Our observation from the other side of it is that the clients who are hardest to serve are the ones sending you incomplete or wrong inputs, then disputing the output. A client whose valuations arrive clean and whose review comes back once instead of three times is a cheaper client to serve. That is a shared interest, and we'd rather talk about it than pretend the two categories overlap. They don't.

Try it on your own documents

The free DDQ tool is the cheapest way to see whether the gap work is automatable at your firm. Bring a real questionnaire and get a draft answer to every question, each with a confidence score. No account required.