Every grant cost transfer checked for allowability, timing and its four answers before it posts, then decided by your reviewers
Late or unallowable transfers are stopped before they post, each with a full file.
See one case, screen by screen ↓Grant Cost Transfer Review checks each cost moved onto a grant for allowability and timing, for post-award staff. A grant cost transfer moves a charge already on the ledger onto a sponsored award; seven agents read the original posting, invoice, order or payroll line, check it against the receiving award and the cost principles, grade the four required answers and look across transfers for costs moving off overspent awards. The department initiates, the post-award reviewer decides with a reason code, and the director gives the second approval on late or large transfers.
AI in education: a buyer's guide to agentic AI for colleges, universities and school districts →
Updated
A grant cost transfer moves a charge that is already on the ledger onto a sponsored award: a supply order keyed to the wrong award, a salary distribution never updated, a recharge posted before the award was set up. Each cost moved onto a federal award must be allowable, allocable to that award, inside its period, timely and explained. A post-award reviewer reads the ERP ledger, the order, the invoice and the receiving award's notice one transfer at a time, and the cost transfer justification often says only "charged in error", which the NIH Grants Policy Statement says is not sufficient on its own.
Some problems show only across transfers. When an award is overspent and about to end, costs drift onto awards with money left; each cost transfer looks small, and the pattern stays hidden from a reviewer reading them one by one. Salary moves after effort is certified. A late cost transfer goes back and forth by email until the second justification is written to fit. When the auditors sample 40 transfers for the Single Audit, the file for each is spread across the ERP, the research administration system, document management and email. Grant Cost Transfer Review puts seven agents in front of that reading, so the reviewer spends her morning on the transfers that need her.
Estimated split of the business days a late, incomplete cost transfer spends from submission to a final decision, by hand and with the agents. Our arithmetic, kept modest. In the working solution the sample transfer was opened at 10:04 a.m., rejected at 10:12 and confirmed by the director at 10:25 the same morning.
Seven agents draw each transfer as an arrow with a 90-day clock, check it against the award and the cost principles, and spot overruns moving; reviewers decide.
On Monday, September 14, at 3:52 p.m., Kyle Brennan initiated CT-26-04417 at Halden State University: $14,860 in lab supplies, moving from NSF 2214587, which ends September 30 and is $15,200 overspent, to NIH R21 ES036118, which has $86,400 left. Dr. Elena Marsh is the investigator on both. Here is what happens to it on Tuesday morning, screen by screen, in the working solution.
Overnight the agents read 41 new transfers from the ERP with their original postings, invoices and payroll lines. Hannah Velez's queue shows 38 waiting, $155,214 proposed to move: 6 past the 90-day window, 8 that need the director too, 4 flagged by the Pattern agent and 23 ready to approve in one pass. CT-26-04417 sits at the top, its ring at 117 days, with all four answers missing and the agents recommending reject at 0.91.
NSF 2214587 · ends Sept 30 · $15,200 overspent → $14,860 lab supplies → NIH R21 ES036118 · $86,400 left
Hannah opens the transfer. The ledger shows PVDF membrane sheets $8,920.00, NMP casting solvent $3,480.00 and stirred filtration cells $2,460.00, posted Thursday, May 21, 2026 to NSF 2214587. Order HS-P2611873 says "membrane casting, Aim 1". The approval steps show nothing posts to the ERP until both the reviewer and the director sign.
The university's cost transfer policy asks four questions: why the cost is moved, how it benefits the receiving award, why it is late and how the error will be prevented. The Justification grader marks all four missing, each with why and the rule cited. The Allowability agent's six checks find the cost inside the R21's period, but no R21 aim uses membrane casting, and the transfer moves 98% of the NSF award's overrun.
Kyle Brennan, Sept 14, 3:52 p.m.: "Charged in error." · 0 of 4 answered
The fan shows $52,322 leaving NSF 2214587 for three accounts, 4 transfers posted and 3 pending, and none in the 11 months before. If the 3 pending ones post, the award goes from $15,200 over to $9,812 under, 15 days before it ends. The agent cites 2 CFR 200.405(c) and proposes three steps; Hannah holds the 2 other pending transfers, asks the director for a look-back on the 4 posted ones, $27,310, and drafts a notice to Dr. Marsh and the department chair.
Same investigator on both sides of 5 of 7 transfers.
Hannah rejects the transfer with the reason code "Moves an overrun or avoids award terms (2 CFR 200.405(c))". Because it is over $10,000 and 117 days late, the rejection waits for the director's second approval before anyone is told or anything changes.
The Ready to approve filter holds 23 transfers, $48,217: each under $10,000 and 90 days, every required answer given, and nothing found by the agents in the ledger, the award or across transfers. Hannah is the second person on each and approves them together. Each posts as its own ERP journal entry with its own audit entry; bulk approval is capped at 25 per action.
Roland's queue shows each reviewer's decision next to what the agents found: 6 late or large transfers waiting, and 1 where the agents disagree with the reviewer, which needs a reason code to confirm. On CT-26-04417 the reviewer and the agents agree, and he confirms. Nothing posts, and Kyle Brennan and Dr. Marsh are told.
The transfer file is sealed with 11 of 11 items: AP voucher 88412-3, the supplier's invoice, purchase order HS-P2611873, proposed journal JE-26-114892, the justification, Dr. Marsh's certification, the agents' checks with their versions, the pattern finding, both decisions and the notice. It is kept 7 years after the award closes, and identifiers are redacted on export.
Kyle starts a $2,318.00 transfer to the R21 in four steps: the cost, the award, the four answers and the investigator's certification. The Justification grader grades each answer as he writes and drafts from the order and lab notebook page 41; why late reads "not needed" on day 5 of 90. It shows 3 of 3 needed answers before he submits, and only the investigator can certify the charge.
Every agent check, decision, refusal, view, setting and export is on one trail, 186,420 entries this fiscal year, with the transfer's history in order: Kyle's initiation, Dr. Marsh's certification, the agents' checks, Hannah's rejection and Roland's confirmation. Walter filters it, exports CSV and builds a redacted export pack for the external auditor, the agency reviewer or the inspector general.
The same grant cost transfer, seen by the four people who carry it, from the department to the auditors: what their day looked like, and what it looks like now.
Seven agents draw each transfer as an arrow with a 90-day clock, check it against the award and the cost principles, and spot overruns moving; reviewers decide.
Reads each proposed transfer from the ERP with its original posting, invoice, order or payroll line, matches the amounts and routes it by department.
Counts the days since the cost first posted, marks day 90, checks the period of performance, pre-award and closeout windows, and routes late or large transfers to the director.
Checks each cost against the receiving award's notice, budget, period and terms, 2 CFR 200.403–405 and the sponsor's policy, and recommends with a confidence.
Grades the four answers, coaches the department administrator while they write, and drafts notes and answers only from the evidence.
Looks across transfers for costs moving off overspent or ending awards, round amounts and repeated moves, and proposes holds, look-backs and notices.
Compares salary transfers with certified effort and payroll, and asks for recertification above the tolerance.
Keeps one file per transfer, seals it when the decision is final and builds audit packs with redaction.
Decide, with reason codes. The agents propose; a named person decides.
Ask in plain words, from the queue or inside a transfer. Every answer points to the order, the award notice, the ledger or the rule it came from.
Why shouldn't CT-26-04417 post?
Three reasons. The order bought the supplies for NSF 2214587, Aim 1. No R21 aim uses membrane casting, so no benefit is shown. And it moves 98% of a $15,200 overrun on an award ending in 15 days. The justification, "charged in error", answers none of the four questions. I recommend rejecting it; Roland Achebe gives the second approval.
Why reject rather than return?
Because no answer could make it allowable as it stands. The order shows the supplies were bought for NSF 2214587, Aim 1, so the original charge looks correct, and moving it would cover that award's overrun, which 2 CFR 200.405(c) does not allow. A return would ask Kyle to write a justification for a transfer that should not happen.
Which awards are losing costs just before they end?
Two. NSF 2214587, ending Sept 30: 7 transfers off it in 3 weeks, $52,322, while $15,200 overspent. NASA 80NSSC24K0391, ending Sept 30: a round $15,000.00 onto it with $31,040 left, while the cited postings total $14,212.40. Moving costs to cover an overrun is not allowed.
What does the Single Audit sample still need?
38 of the 40 sampled files are complete. CT-25-11804 is missing the PI certification and CT-25-12960 the invoice; I can request both from the departments. Last year's finding was 9 of 40 late without explanation.
13 screens from the working solution, on its sample data. Pick one to see it large.
The 90-day window, the thresholds, the response times and retention are settings. Thresholds belong to the director; every change is saved as a new version and logged.
| Setting | Default | Choose from |
|---|---|---|
| A transfer is late after | 90 days | Counted from the day the cost first posted; late transfers must answer why late and how it will be prevented |
| Second approval by the director | Over $10,000 | Also for every late transfer and for rejections, each on or off; set by the director only |
| Reviewer and director decide within | 5 and 2 business days | Escalate overdue work after 2 business days to the director, then the Associate Vice President for Research Finance |
| Bulk approval, at most per action | 25 transfers | Only clean transfers under every threshold |
| Refuse "charged in error" and "correct project" as a reason | On | On · Off |
| Transfers onto an award after it ends | 120 days | Only for costs incurred inside its period, until closeout; pre-award costs up to 90 days before the start where the sponsor allows |
| Effort change that needs recertification | 5 points | Salary transfers above it are held until the investigator recertifies |
| Keep transfer files after the award closes | 7 years | Federal minimum 3 years from the final financial report; legal hold suspends deletion |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
Sources: NIH Grants Policy Statement 7.5 (cost transfers; "to correct error" is not sufficient; transfers to cover overruns are not allowable) · 2 CFR 200.403, 200.405, 200.430 and 200.334 · OMB Compliance Supplement, Part 3 · NSF NCSES, Higher Education Research and Development Survey, FY2024 ($64 billion federal) · "target" = design goal, measured in the live solution · "demo" = seen in the working solution, on its sample data · "typical" = published figures · "estimated" = our arithmetic, kept modest. Halden State University, its awards, suppliers, transfers and the people named on this page are fictional: characters and sample data in the working solution.
Grant Cost Transfer Review checks each cost moved onto a grant for allowability and timing, for post-award staff. A grant cost transfer moves a charge already on the ledger onto a sponsored award; seven agents read the original posting, invoice, order or payroll line, check it against the receiving award and the cost principles, grade the four required answers and look across transfers for costs moving off overspent awards. The department initiates, the post-award reviewer decides with a reason code, and the director gives the second approval on late or large transfers.
It is the department's written explanation of a transfer. In the working solution, as in the university's cost transfer policy, it answers four questions: why the cost is moved, how it benefits the receiving award, why it is late and how the error will be prevented. The NIH Grants Policy Statement says an explanation that merely states "to correct error" or "to transfer to correct project" is not sufficient, so the Justification grader refuses "charged in error" and coaches the department administrator before the transfer is sent.
In the working solution, a transfer made more than 90 days after the cost first posted is late, as the university's cost transfer policy sets it; the window is a setting. A late transfer must also answer why it is late and how the error will be prevented, it needs the director's second approval, and its ring on the reviewer's queue turns amber at day 75 and red past day 90.
Under the federal cost principles a cost must be reasonable, allocable to the award and consistent with its terms. 2 CFR 200.405(a) allocates a cost by the benefit the award receives, and 200.405(c) bars moving costs to another federal award to overcome fund deficiencies. The Allowability agent checks each transfer against the receiving award's notice, budget, period and terms, and cites the rule for every finding.
The Effort agent compares the payroll lines being moved with the certified effort report. In the working solution, an $8,450 salary transfer would change certified effort from 50 / 25 / 25 to 35 / 40 / 25, above the 5-point tolerance, so the agent recommends a hold and the reviewer asks the investigator to recertify.
No. The agents read, check, grade, flag and draft; they never approve, reject, post a journal, hold a transfer, open a look-back or send a notice. Below each agent's confidence threshold the transfer goes to a person, and a person who overrides an agent gives a reason code.
Yes. The department administrator initiates and the investigator certifies; the post-award reviewer approves, returns or rejects; the director gives the second approval over $10,000, on every late transfer and on rejections. Nobody decides a transfer they initiated or certified, and every refused action is logged.
The Agentic Solution Engine builds and deploys it from your requirements and documents: your cost transfer policy, your thresholds and a sample of decided transfers. It goes live once every quality gate has passed, and we will walk you through it on your own transfers first.
Five high-fit calls in minutes, each limit and cost share cited to the sponsor's own text.
1-min demoGrant Proposal ReviewEvery proposal reviewed inside the five-day window, every value tied to its clause.
1-min demoSubrecipient Risk Assessment & MonitoringRisk scored before signing and every invoice line checked before the university pays.
1-min demoWe'll run Grant Cost Transfer Review on a sample of your own cost transfers and your cost transfer policy.