Fund administration oversight that grades every administrator on its evidence, not on its own SLA report
Every administrator graded on its evidence, with the cause of each error found and fixed.
See one case, screen by screen ↓Fund administration oversight is how an asset manager shows it oversees the firm that keeps its funds' books: service levels measured, NAV errors found and fixed, the administrator's SOC 1 report read, its own complementary controls run, and a due diligence review every year. Fund Administration Oversight gives each administrator a quarterly report card, graded by seven agents on the administrator's own deliverables against its contract, proposed by an analyst and signed by the fund COO.
AI in asset management: a buyer's guide to agentic AI for investment managers →
Updated
Outsourcing a fund's books does not outsource the NAV. Under Rule 38a-1 a registered fund's compliance policies have to provide for oversight of compliance by its administrator, and the CCO reports on it to the board every year; AIFMD II asks every AIFM to describe its periodic due diligence on delegates. Yet outsourcing oversight often comes down to a quarterly scorecard rebuilt by hand from the administrator's own monthly SLA report, which can measure against the administrator's own clock and leave out days it chose to exclude.
The SOC 1 Type 2 report runs to well over a hundred pages and is read once and filed, so nobody links an exception deep inside it to the NAV restrikes the funds actually paid for. The complementary user entity controls the manager must run itself rarely have an owner or evidence. And the annual due diligence questionnaire asks the administrator two hundred questions or more, many of which the manager could answer from the year's own records. Errors repeat because their control cause is never named.
Estimated split for one large administrator over one quarter, by hand and with the solution. The annual due diligence review comes on top: an estimated 4–6 weeks by hand, a target of 3 days with the agents.
Seven agents collect, measure, grade, challenge and draft; the analyst proposes, the fund COO signs.
Quillfield Capital has outsourced 180 funds, $745.2B of assets, to six fund administrators. It is Monday, October 19, 2026; the vendor oversight committee meets on October 29. Pellmont Fund Services, which runs 34 of Quillfield's US fixed income and multi-asset funds, has had a bad quarter. Here is how fund administration oversight works in the working solution, screen by screen.
Lena's home screen shows 2 of 6 report cards approved, due by October 21, 9 open SOC 1 exceptions and 11 remediation commitments, 2 overdue. Under “Needs you” sit five decisions, one next step each. Beside them she can ask the agents anything — “What needs me before the committee?”, “Why is Pellmont a C?” — and every answer cites the record.
Grading agent, 8:31 a.m.: “Pellmont drops to C: seven restrikes, four from the stale-price weakness its own auditor found. Its card and letter are ready for you.”
Every administrator has a grade for timeliness, accuracy, controls, incidents, responsiveness and staffing, an overall grade and a status. Harrowgate and Tessaly are approved. Pellmont is a C, awaiting her approval. Corbière shows A/B because the agents disagree, and Aldermoor is I, incomplete, because its SOC 1 report is overdue.
“Proposed by the analysts, approved by the Fund COO.”
Pellmont's report card reads like a school report: four terms across, the Q3 grade with its evidence count, and a comment from the agents. Clicking the C in Accuracy opens the thread: the Evidence agent found 7 NAV restrikes in 4 funds, $186,400 borne by the funds, two over $0.01 per share. The Grading agent proposes C under rubric §2.2. The Challenge agent re-grades blind on a second model and agrees.
Evidence agent: “None reached 0.5% of NAV, so no shareholder reprocessing was needed.”
The Intake agent logged 1,488 Pellmont deliverables — NAV packs, reconciliations, pricing exception reports, SLA reports — and timestamped each on receipt in Quillfield's systems. Against the 98.5% target in Schedule 3, 1,431 arrived on time: 96.2%, with 51 late and 6 missing. Pellmont's own SLA report says 99.1% because it measures against 8:00 p.m. and leaves out August 12. The contract says 6:30 p.m. and excludes nothing.
“That is why we grade from our timestamps, not their report.”
The Controls agent read Pellmont's SOC 1 Type 2 report from Ashby Lorne LLP: unqualified opinion, two exceptions, a bridge letter covering July 1 to September 30. Control 6.2 on page 88 — the daily stale-price review — had no evidence of review for 3 of 40 days sampled. The agent linked it to the four stale-price restrikes Quillfield saw: RS-0711, RS-0722, RS-0909 and RS-0923.
“Four restrikes with the same cause as this exception: stale prices that nobody investigated. This is why Controls is a C, not a B.”
Priya Raman, the oversight analyst, proposed the card on Friday, so Priya cannot approve it; the button says why. Lena reads the restrikes and page 88, then approves. Two grades are below B, so a reason code is required: she picks GRADE-EVIDENCED and signs. The card and its evidence are frozen and go to the vendor oversight committee pack.
“I, Lena Fischer, approve this report card for Q3 2026. Electronic signature; re-authenticated with single sign-on.”
The Remediation agent drafted the letter to Pellmont's client service director, citing all seven restrikes, the 96.2% against 98.5% and SOC 1 page 88, with four commitments by date: daily stale-price review evidence and a second price source for any bond unchanged five or more business days among them. It computed a $186,400 service credit under §3.6. Lena signs the letter and approves the credit; Daniel Okafor, the vendor management lead, sends it and claims the credit.
“The agent computes; it never claims. The person who approves cannot claim, and the person who claims cannot approve.”
For Corbière's Luxembourg and Irish sub-funds, the Grading agent gave Timeliness an A: 98.8% against the 10:00 CET cut-off in the SLA catalog. The Challenge agent found Avenant n° 2, a French amendment in force from April 1 that moved the ICAV cut-off to 08:30 CET. On that clock nine more packs were late: 98.1%, a B. Both readings sit side by side with their sources, and Aoife decides.
Grading agent: “This is a question about the contract, so a person decides.”
Pellmont's 2026 review is drafted from the year's evidence. Of the 212 questions in Quillfield's DDQ, 187 are answered from deliverables, restrikes, incidents, the SOC 1 report and staffing records, and 19 cite Pellmont's own response. Only 6 go to Pellmont, such as whether it takes a second price source for securities whose price has not changed for five or more business days. The proposed conclusion — retain, on watch — waits for Lena.
“Pellmont is not asked the other 206 questions again.”
The same quarter of fund administration oversight, seen by five of the people who carry it — what it looked like, and what it looks like now.
Seven agents collect, measure, grade, challenge and draft; the analyst proposes, the fund COO signs.
Collects NAV packs, reconciliations, SLA reports, incident notices, SOC 1 reports, bridge letters and DDQ responses from each administrator's portal or SFTP, timestamps each against its contractual cut-off, and flags and chases late and missing items.
Measures each subject from the records of truth: on-time rate, restrikes and NAV errors, breaks over tolerance, incidents by severity, query turnaround, staffing turnover and key-person changes.
Proposes a letter grade per subject by the firm's written, versioned rubric, cites the clause that set it, and writes the card's comment.
Re-grades every subject blind from the same evidence on an independent model, searches for contract amendments in any language, and escalates disagreements with both readings.
Reads SOC 1 Type 2 and ISAE 3402 reports and bridge letters — opinion, period, exceptions with their page, management responses, subservice organizations — links exceptions to errors the firm saw, and maps complementary user entity controls to owners and evidence.
Drafts the annual review: answers the firm's DDQ from the year's evidence, cites the administrator's own response, lists the gaps and drafts a proposed conclusion.
Drafts remediation letters citing each error and exception, computes service credits under the SLA schedule, tracks commitments and checks the evidence an administrator uploads against each one.
Maker, approver, sender, co-signer. The agents propose; a named person decides.
Every working screen has an ask box. Answers cite the record, and the agents act only when you approve.
Why is Pellmont a C?
Accuracy and Controls are both C, and rubric §2.7 takes the lower of the two when either is C. Accuracy: seven restrikes, two over $0.01 per share. Controls: SOC 1 exception 6.2 (stale-price review, p. 88) is tied to four of those restrikes. The Challenge agent agreed blind on all six subjects.
Why does their SLA report say 99.1%?
Pellmont measures against 8:00 p.m. and excludes Aug 12 as a third-party outage. The agreement says 6:30 p.m. and has no outage exclusion; receipt is measured in our systems. Of the 51 late items, 38 arrived before 8:00 p.m.
Is part of this weakness ours?
Partly. The report's complementary control 4 says we review and approve stale-price challenges Pellmont raises. For August, 9 challenges were raised and our approvals were not retained. That gap is on our attestation, not in Pellmont's grade.
Does anything here move money?
One item: the $186,400 service credit under SLA §3.6, itemized by restrike. You approve it; Daniel Okafor (vendor management) claims it against Pellmont's Q4 invoice. Nobody can do both, and I never claim it myself.
12 screens from the working solution, on its sample data. Pick one to see it large.
The rubric, the SLA catalog and the clocks are settings. Every change needs a second person and takes effect from the next term.
| Setting | Default | Choose from |
|---|---|---|
| Timeliness, grade A | ≥ 98.5% on time against the contract cut-off | B ≥ 96% · C ≥ 93% · D below 93% |
| Accuracy, grade C | 3–7 restrikes, or any of $0.01 per share or more | A no restrikes · B 1–2, none of $0.01 · D more than 7, or any of 0.5% of NAV |
| Controls, grade C | An exception tied to an observed error, or 4 or more | A unqualified, no exceptions · B 1–3 not tied to an error · D qualified opinion · I report overdue |
| Overall grade | Lowest of Accuracy and Controls when either is C or below | Otherwise the rounded average · overall C puts the administrator on watch · two Cs in a row or any D escalates to the committee |
| Agents propose a grade at | Confidence ≥ 0.90 (Grading, Challenge) | Controls ≥ 0.85 · below the threshold, a person reads the evidence first |
| Report cards approved by | Business day 15 after quarter end | Proposed by business day 10 · escalate to the CCO at + 2 |
| Remediation commitments | Reminder 5 business days before due | Vendor lead at due + 1 · COO at + 5 |
| Records and residency | Kept 7 years, primary region US East | EU copy for Luxembourg and Irish fund records · Asia copy in Singapore for Singapore and Japan fund records · legal hold can be set per administrator or fund |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal, measured in the live solution · “estimated” = our estimate, including the 4–6 week due diligence baseline · “typical” = published figures (The Drawdown, Fund admin report 2025, July 23, 2025 — $2.5 trillion of assets under administration and almost 3,000 funds added by the 12 largest administrators; SEC order against an investment adviser, October 18, 2016, File No. 3-17630 — $3.9 million penalty, about $27 million contributed to the funds). Regulations named as context: SEC Rule 38a-1; Directive (EU) 2024/927 (AIFMD II), Art. 7(2)(e)(iv). People, firms, administrators, funds and service auditors named on this page are fictional — characters and sample data in the working solution.
Fund administration oversight is how an asset manager shows it oversees the firm that keeps its funds' books: service levels measured, NAV errors found and fixed, the administrator's SOC 1 report read, its own complementary controls run, and a due diligence review every year. Fund Administration Oversight gives each administrator a quarterly report card, graded by seven agents on the administrator's own deliverables against its contract, proposed by an analyst and signed by the fund COO.
Yes. Rule 38a-1 requires a registered fund's compliance policies to provide for oversight of compliance by its administrator, along with its adviser, principal underwriter and transfer agent, and the CCO's annual written report to the board addresses each one. That makes service provider oversight a standing part of the fund's compliance program. In the solution the board's annual service-provider review is drafted from approved report cards only, and the CCO edits and signs it. For AIFMs, AIFMD II asks for a description of the periodic due diligence on each delegate.
In Fund Administration Oversight, the Controls agent reads each SOC 1 Type 2 or ISAE 3402 report and its bridge letter for the opinion, the period, every exception with its page, management's responses and the subservice organizations. It then links each exception to the restrikes, breaks and incidents the firm actually saw. An overdue report puts Controls at I, incomplete, and is chased: a missing report never counts as clean.
Complementary user entity controls are the controls the manager itself must run for the administrator's controls to work, such as reviewing the stale-price challenges the administrator raises. The Controls agent maps every one from each SOC 1 report to an owner and the quarter's evidence — 61 in the sample firm, 57 evidenced and 4 gaps — and the fund COO and the CCO both sign the quarterly attestation.
Yes. The Due diligence agent answers the firm's DDQ from the year's deliverables, restrikes, incidents, SOC 1 report and staffing records, and cites the administrator's own response where that is the only source. In the sample review it answered 187 of 212 questions and left 6 real gaps for the administrator. It never sends a question on its own, and the COO approves the conclusion.
Because the administrator's own report can use its own clock. In the sample quarter Pellmont's report said 99.1% on time against 8:00 p.m. and left out an outage day; measured on receipt against the contract's 6:30 p.m. cut-off, with no exclusion, it was 96.2%. Only an analyst can excuse a late deliverable, with a reason.
Yes. The agents measure, propose, explain and draft; they never set a final grade, excuse a delay, send a letter, claim a credit, close a commitment or attest a control. The analyst proposes, the fund COO approves with a reason code and an electronic signature, and money needs two people: the COO approves a service credit and the vendor management lead claims it.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your administration agreements and SLA schedules, your grading rubric, your DDQ and a recent quarter of deliverables and SOC 1 reports — and it goes live once every quality gate has passed. We will walk you through it on your own administrators first.
Every NAV checked against your books before it goes out, with breaks held, not published.
1-min demoFund Expense AllocationEvery invoice split with its LPA clause, and nothing the manager owes charged to a fund.
1-min demoManagement Fee CalculationEvery fee recalculated from its clause before billing, every gap explained and signed.
1-min demoWe'll run Fund Administration Oversight on a quarter of your own administrators' deliverables and SOC 1 reports.