Single audit preparation that runs the mock audit first, with every sample cited and every finding signed before fieldwork
Repeat findings caught two months before fieldwork, each with a signed plan.
See one case, screen by screen ↓Single Audit Preparation is a set of agents that prepares a college or university for its single audit by running a mock audit before fieldwork. Seven agents build the schedule of expenditures of federal awards and tie it to the ledger, work out the expected major programs, test samples the way the external auditor will, and draft findings and corrective action plans. Grants accountants confirm what clears, and the Associate Controller signs every finding and every package before the auditor sees it.
AI in education: a buyer's guide to agentic AI for colleges, universities and school districts →
Updated
Single audit preparation starts when the fiscal year closes: every university, college and school district that spends $1,000,000 or more of federal awards in a year gets a Single Audit under 2 CFR 200 Subpart F. The controller's office pulls an ERP report by fund and fixes it by hand into a schedule of expenditures of federal awards: Assistance Listing numbers mapped wrong, cost share sitting in sponsored funds, Direct Loans and Perkins balances that are not in the ledger, pass-through identifiers missing. Then the auditor's request list arrives, a few hundred prepared-by-client items, and staff email departments and export reports from the grants system, effort reporting, payroll and the financial aid system.
Nobody tests the way the auditor tests until the auditor arrives. Exceptions surface at fieldwork, when the only remedy left is the corrective action plan, and the same effort, procurement, period-of-performance and enrollment-reporting findings come back year after year. A material weakness in a major program costs the institution its low-risk auditee status for two years. Single Audit Preparation runs the audit first: an auditor agent tests samples against the Compliance Supplement, a preparer agent answers from the systems of record, and the Associate Controller reads and signs the findings two months before the auditor writes them.
Estimated split of the finance office's staff-days for one audit year, by hand and with the agents; our arithmetic, kept modest. The time to assemble request-list evidence is a design target, measured in a pilot.
An auditor agent tests the way the external auditor will; a preparer agent answers from the systems of record; the Associate Controller signs.
Halden State University, fiscal year July 1, 2025 to June 30, 2026. It is Tuesday, September 15, 2026, and the external auditor, Ashford & Vale LLP, starts fieldwork on Monday, November 16. Here is the mock audit, screen by screen, in the working solution.
Priya opens the mock audit. The federal awards schedule sits in the top strip: $968,042,615 across 2,140 awards, tied to the ledger, with the three major clusters named and 62 days to fieldwork. Eight samples are queued for today's run, and the 420-sample plan shows what has cleared, what needs a person and what became a draft finding.
Fieldwork begins Monday, November 16 — 62 days
The ERP's own report said $970,359,495. The Schedule agent started from $695,440,495 in the ledger, took out $2,316,880 of institutional cost share and added $268,904,000 of Direct Loans and $6,015,000 of Perkins balances: difference $0. Each reconciling item became a cited adjustment, SA-01 to SA-03, and Priya approved all three on September 11.
With $968.0M of federal spending, the Type A threshold is $3,000,000. Research and Development and Student Financial Assistance had findings last year that are not yet corrected, and TRIO was not audited as major in FY2024 or FY2025, so all three are high risk. Coverage is 98.9% against the 20% a low-risk auditee needs.
The external auditor decides; this is what to plan for.
The auditor agent runs the period-of-performance test on sample 23, $48,200 of equipment on award R01NS124417. The purchase order is dated July 11, 2025, and the ERP shows the award ending June 30, 2025, eleven days earlier. On the timeline, the purchase sits past the ERP date.
The purchase order is dated July 11, 2025; the ERP shows the award ending June 30, 2025 — 11 days earlier.
The preparer agent looks beyond the ERP and finds, in the grants system, a notice of no-cost extension sent June 12, 2025, with a revised end date of June 30, 2026. The purchase is inside the period, and the sample clears at 97% against a 95% threshold. The agent opens follow-up FU-118 to correct the ERP end date, and Priya can open the cited passage herself.
Sample 41 is $38,412 of a postdoctoral researcher's salary at 60% on award CHE-2318406, July to December 2025. The effort report matches payroll but was never certified: it was due March 2, 2026 under policy FIN-412, and reminders on March 9, April 6 and May 4 went unanswered. With samples 133 and 247, 3 of 60 compensation samples are unsigned, $71,906 in all.
The Finding drafter writes DF-01, proposed number 2026-001, in the auditor's five-part form with $71,906 of known questioned costs, every sentence cited. It marks the finding as a repeat of 2025-003, when 4 of 60 were late, and says why last year's corrective action did not hold: reminders without escalation. The plan adds what was missing, with three actions, three owners and dates through December 31, 2026.
Renata signs DF-01 with a management position code, SGN-01 agrees, SGN-02 agrees in part or SGN-03 disagrees, writes the response for the auditor, and confirms she read the evidence for samples 41, 133 and 247. Priya, who prepared it, cannot sign.
DF-01 is signed on Tuesday, September 15, 2026, 62 days before fieldwork begins. The signature, the reason code and the evidence go on the audit trail, and the corrective action plan is in place before the auditor draws a single sample.
Nadia reads, filters and exports; she changes nothing. The trail shows Renata signing DF-01, the Finding drafter writing it and the auditor agent clearing sample 23 from the extension notice. One export carries the schedule, the major-program workpaper, all 420 sample results, the signed findings, the release log and the agent versions.
The same audit, seen by the four people who carry it: what their autumn looked like, and what it looks like now.
An auditor agent tests the way the external auditor will; a preparer agent answers from the systems of record; the Associate Controller signs.
Builds the federal awards schedule by cluster, agency and Assistance Listing number from the ERP, the grants system and the federal loan record, ties it to the ledger and explains every difference.
Works out the Type A threshold, the risk of each program, any Type B assessment and the coverage, as the external auditor will.
Draws samples the way the external auditor does, random with a recorded seed plus every high-dollar item, runs the Compliance Supplement test steps and challenges the evidence.
Answers each test step from the ERP, the grants system, effort reporting, payroll, the financial aid system and the document store, attaching the documents with the passage cited.
Writes failed tests as findings in the five-part form, compares them with the last three years of findings, says why a repeat repeated, and drafts the corrective action plan.
Assembles each package on the auditor's request list, indexes and redacts it, reminds owners and escalates past-due requests.
Checks daily for a new Compliance Supplement and, when one is issued, lists every changed test step for the Associate Controller to approve.
Confirm, sign, release. The agents propose; a named person decides.
The assistant answers in plain words from the same evidence the agents used, with every answer cited to its source.
Why did sample 23 clear?
The purchase order is dated July 11, 2025 and the ERP shows the award ending June 30, 2025. The grants system has a no-cost extension notice sent June 12, 2025 moving the end to June 30, 2026, sent 18 days ahead, inside the 10-day rule. The purchase is in the period. I opened follow-up FU-118 to correct the ERP date.
Which findings repeat last year?
Two. DF-01, unsigned effort certifications, repeats 2025-003: 3 of 60 samples, $71,906. DF-05, enrollment changes reported late, repeats 2025-001. Finding 2025-002 on subrecipient risk assessments is fully corrected: 40 of 40 this year.
How were the major programs chosen?
Federal spending is $968,042,615, so the Type A threshold is $3,000,000. Research and Development and Student Financial Assistance had findings last year that are not yet corrected, and TRIO was not audited as major in either of the last two years, so all three are high risk. Coverage is 98.9% against the 20% a low-risk auditee needs. The external auditor decides; this is what to plan for.
Draw 10 more procurement samples over $350,000
Drawn with the same seed, 20260901-HSU-FY26: 10 purchases of $350,000 or more under requirement I, from 64 in the population. They are queued for tomorrow's 07:00 run and listed under Samples. The draw is on the audit trail.
13 screens from the working solution, on its sample data. Pick one to see it large.
How samples are drawn, when an agent's clear needs a person and how student records are handled are settings, versioned and on the audit trail.
| Setting | Default | Choose from |
|---|---|---|
| Random seed | 20260901-HSU-FY26, recorded | Re-drawing with the same seed returns the same items |
| High-dollar amount | $100,000 | $50,000 · $100,000 · $250,000 |
| Sample size table | 25 · 40 · 60 | By population and assessed risk, as the external auditor sizes samples |
| Clear threshold | 95% | 90% · 95% · 98%; a person confirms every agent clear |
| Compliance Supplement edition | 2025 in use | Changed test steps approved before any retest |
| Student records | Masked; reveal with a reason | Kept seven years; the federal minimum is three after the report is accepted |
| Clocks | Needs you: 3 business days | Packages signed 5 business days before due; past-due escalation after 3 |
| Agent and model versions | Frozen August 20 | Changes go to the administrator and the Associate Controller |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
Sources: 2 CFR 200.501, 200.511, 200.512, 200.518 and 200.520 (Uniform Guidance, as revised for fiscal years beginning on or after October 1, 2024) · 2 CFR 200.308(e)(2) (no-cost extensions) · 34 CFR 668.22 (return of Title IV funds, 45 days) · a state legislative auditor's 2023 statewide Single Audit (18 of 31 findings repeated) · "target" = design goal, measured in a pilot · "demo" = seen in the working solution, on its sample data · "typical" = published figures and rules · "estimated" = our arithmetic, kept modest. Halden State University, Ashford & Vale LLP and the people named on this page are fictional: characters and sample data in the working solution.
Single Audit Preparation is a set of agents that prepares a college or university for its single audit by running a mock audit before fieldwork. Seven agents build the schedule of expenditures of federal awards and tie it to the ledger, work out the expected major programs, test samples the way the external auditor will, and draft findings and corrective action plans. Grants accountants confirm what clears, and the Associate Controller signs every finding and every package before the auditor sees it.
A Single Audit is the audit of federal awards required by 2 CFR 200 Subpart F. For fiscal years beginning on or after October 1, 2024, it applies to any non-federal entity that spends $1,000,000 or more of federal awards in a year; the threshold was $750,000. The reporting package is due no later than nine months after the fiscal year ends.
Under 2 CFR 200.511(c), the plan names the corrective action, the contact person and the completion date for each finding. In the working solution, the plan for DF-01 has three actions, three owners and dates through December 31, 2026, and adds the escalation that last year's reminders-only plan lacked.
The Major Program agent applies 2 CFR 200.518 to 200.520 to the signed schedule and the last two years of findings: the Type A threshold, each program's risk with its reason, any Type B assessment and the coverage. It shows what would change the determination, such as a material weakness removing low-risk status and raising coverage from 20% to 40%. The external auditor decides; this is what to plan for.
The Supplement agent checks daily for a new edition. When one is issued, it compares it with the edition in use, requirement by requirement, and lists every changed test step for the Associate Controller to approve before any sample is retested. In the working solution, testing stays on the 2025 edition because the 2026 edition has not been issued.
Yes. The agents test, answer and draft; they never post an entry, edit a source record, send a finding or release a package. A person confirms every agent clear, samples below the threshold go to a named person, and only the Associate Controller can sign a finding, its plan, the schedule or a request package. Only the audit liaison can release a signed package.
Student identifiers are masked throughout, including in request packages. Only the Director of Financial Aid or the auditor can reveal a student, with a reason, for one sample, and every reveal is logged. Records are kept seven years, then deleted.
The Agentic Solution Engine builds and deploys it from your requirements and documents, such as your award list, last year's findings and the auditor's request list, and it goes live once every quality gate has passed. We will walk you through it on your own awards first.
Every federally funded buy documented the day it is requested, with its proof.
1-min demoSection 117 Foreign Gift ReportingNo foreign source missed or understated, every reported line traced to its record.
1-min demoFacility Condition AssessmentCapital dollars go to the riskiest systems first, and every rank shows its proof.
1-min demoWe'll run Single Audit Preparation on a sample of your own awards, ledger and last year's findings.