Credit risk review of every commercial loan, every quarter — graded from the file, cited to the page
Every loan reviewed every quarter, not a yearly sample.
See one case, screen by screen ↓Software for a bank’s loan review team — the independent check on the grades lenders give their own loans. Credit Risk Review re-grades every commercial loan from its file each quarter, shows every loan whose grade of record the file no longer supports, and carries each one through the reviewer’s decision, the challenge to the lender and the board report.
How to choose AI solutions for lending →
Updated
Loan review reads a risk-based sample once a year: the large relationships, the watch list, insider loans and a slice of the rest. Every other loan carries the grade the lender gave it until its turn comes round — so a loan graded pass outside the sample can stay mis-rated until a borrower misses a payment or an examiner finds it.
The evidence is usually already in the file. A compliance certificate adds back an expense the credit agreement does not allow, distributions are left off the certificate, a call report mentions the largest customer leaving. Nobody re-reads it, because a full-file review of a mid-size relationship takes a reviewer hours — most of them spent pulling the file together and spreading statements, and little of it spent judging the credit.
Estimated split for one mid-size commercial relationship, in hours, by hand and with the solution.
Six specialist agents re-grade every loan from its file; a loan review officer decides and the lender answers.
Monday 12 October, ten days before the board risk committee. The agents finished reading Northmere Bank & Trust’s whole commercial book on Sunday morning. Here is what Nora Lindqvist does with it, screen by screen, in the working solution.
The grade migration matrix puts each loan’s Q2 grade of record against the grade its file supports in Q3. 1,327 loans sit on the diagonal: the agents agree with the lender. 91 loans, $518.7M, sit off it — 68 graded too generously, 23 too harshly, and 20 pass loans that cross the dashed line into criticized.
How this matrix was made: “1,418 loans re-graded on 8 factors, every reason cited to a page · 1,327 agree with the grade of record · 91 do not.”
Switched to “Last year’s sample plan”, the matrix shows what the old method would have seen this quarter: about 78 loans read — the large ones, the watch list and a slice of the rest. It would have reached 27 of the 91 by the committee. The other 64, $329.1M, keep the lender’s grade until their annual review.
“The other 64 ($329.1M) keep the lender’s grade until their annual review, including all 6 in Pass 4 → Substandard 7.”
The only three-notch jump in the book. The cell opens to its loans with the reason for each move and the page behind it. Brennick Cold Storage, $17.6M, heads the list: debt service coverage 0.84x on the agreement’s terms against a 1.25x covenant — the certificate says 1.31x — and its largest customer moves 40% of its volume in April 2027.
Nora starts the review. The file assembler gathers the credit file and flags an appraisal 40 months old. The financial spreader finds the $680,000 add-back has no written approval under §1.1 and $710,000 paid to the members is missing from the certificate. The covenant tester recomputes 8 covenants and finds two breaches. The risk rating assessor reads the memo and call reports and suggests 7 · Substandard.
“Why not 8 · Doubtful: payments are current, the buildings cover the loan at 67% loan-to-value and the guarantor holds $2.9M of cash — full collection is not improbable.”
Repayment capacity moves from 4 to 7, customer concentration and liquidity to 6, leverage and management to 5. Collateral, the guarantor and payment performance hold. Each row shows what the file says now against renewal, with links to the certificate, the agreement, the Fernvale letter, the call report and the borrowing base — 16 citations in all.
“No better than repayment capacity unless the collateral alone repays the loan (policy 3.2.4).”
EBITDA was $4.715M as reported. Take out the $680,000 compressor add-back the bank never approved, the $850,000 of unfinanced capital spending and the $710,000 of distributions, and $2,475,000 is left against $2,950,000 of debt service: 0.84x, down from 1.48x at renewal. Funded debt to EBITDA is 4.11x against a 4.25x maximum.
Credit agreement §1.1: add-backs only for charges “approved in writing by the Bank in its sole discretion” — no approval in 34 documents.
Eight financial and reporting covenants for the quarter to 30 June. Two breached: debt service coverage (§7.1) and distributions (§7.5), which are allowed only while coverage stays at or above 1.35x. Two tight: leverage and the revolver at 94% of its borrowing base. The signed certificate sits beside the table with the add-back highlighted.
The challenge drafter writes it in the bank’s format: three reasons with six citations, why not Doubtful, and four conditions — a new appraisal of both buildings within 60 days, a reservation-of-rights letter and no further distributions, the borrower’s plan to replace the Fernvale volume, and a monthly borrowing base. Nora can edit every line before she sends it.
The challenge drafter never sends a challenge — the reviewer does.
Marcus has until Mon 26 Oct, 10 business days, to agree or answer with his reasons. He agrees with 7, tells the borrower there will be no more distributions, asks legal for the reservation-of-rights letter and orders the appraisal. Had he disagreed, both write-ups would go to Warren Pike, the Chief Credit Officer.
“I knew about Fernvale but expected Owen to backfill the space — he hasn’t signed anyone yet. I didn’t know about the distributions.”
Nora records the final grade 7. The grade of record changes from 4, and the allowance team and the watch list are told. In Adaeze Okafor’s Q3 loan review report for the board risk committee, Brennick is one of the 20 pass loans newly criticized, $152.7M, next to the review status of all 91 and each lender’s grade accuracy.
The same quarter, seen by the five people who carry it — what their review looked like, and what it looks like now.
Six specialist agents re-grade every loan from its file; a loan review officer decides and the lender answers.
Gathers each loan’s approval memo, agreement, statements, certificates, call reports and core-system data from imaging, the core system and the relationship file, and checks what is missing and what is stale.
Spreads audited, reviewed and interim statements and tax returns, and recomputes EBITDA, coverage, leverage and liquidity on the credit agreement’s own definitions.
Recomputes every financial and reporting covenant from the spread and compares it with the borrower’s compliance certificate: certificate value, recomputed value, headroom and result.
Scores the eight factors of the bank’s 1–9 rating scale, reads the memo and call reports for what the numbers do not show, and suggests a grade with every reason cited to a page.
Ranks every disagreement by notches times commitment, weighting moves into criticized, and assigns each loan to an independent reviewer.
Drafts the grade challenge to the lender — review grade, cited reasons, why not the next grade, conditions with owners and dates — and the summary in the quarterly board report.
Decides; the lender answers. The agents propose; a named person decides.
Reviewers can ask in plain words about any loan, grade, factor or covenant — or tell it what to draft. Every answer points to the page it rests on.
Why do the agents grade Brennick a 7, not a 4?
Three things changed since the renewal at 4: debt service coverage fell from 1.48x to 0.84x; the largest customer moves 40% of its volume in April 2027, though the memo says its contract runs “through 2029”; and the revolver is 92% drawn against 41%. Cash flow does not cover debt service — a well-defined weakness, so Substandard. It is not an 8: payments are current and the buildings cover the loan at 67% loan-to-value.
Which of these grades would last year’s sample have missed?
Last year’s plan reads about 78 loans a quarter. By the committee on 22 Oct it would have read 27 of the 91 loans whose grade is off. The other 64, $329.1M, would keep the lender’s grade until their annual review — including all 6 in Pass 4 → Substandard 7. Brennick Cold Storage was last reviewed in February; its next scheduled review is February 2027.
Is the 1.31x on Brennick’s certificate right?
Not on the agreement’s terms. It adds back $680,000 for the Building B compressors as non-recurring; §1.1 allows that only with the bank’s written approval, and none is in the 34 documents. It also leaves out $710,000 paid to the members, which §7.1 deducts. That gives 0.84x against a 1.25x minimum — a breach.
Draft the challenge to Marcus Hale with a new appraisal as a condition
Drafted at 7 · Substandard with three cited reasons — coverage, the Fernvale notice and the revolver. Condition 1 is a new appraisal of both buildings within 60 days: the June 2023 valuation is 40 months old and policy 3.4.2 requires one for a criticized loan. It is open for you to edit and send.
13 screens from the working solution, on its sample data. Pick one to see it large.
The rating scale, the review cycle, the challenge clock and the independence rules are settings. Every change is versioned and on the audit trail.
| Setting | Default | Choose from |
|---|---|---|
| Rating scale | 1–9, credit policy 3.2 | 1–4 Pass · 5 Watch · 6 Special Mention · 7 Substandard · 8 Doubtful · 9 Loss |
| Read every commercial loan | Quarterly + on new documents | Quarterly + on new documents · Monthly + on new documents · Quarterly only |
| Disagreement that goes to a reviewer | 1 notch | 1 · 2 · 3 notches; any move from pass into special mention or worse always goes |
| Always reviewed by a person | Over $10M, insiders, watch list | Over $5M · $10M · $15M, insiders, watch list |
| Lender response clock | 10 business days | 5 · 10 · 15 business days |
| When the lender disagrees | Warren Pike, Chief Credit Officer | Chief Credit Officer · Credit committee |
| Re-appraisal for a criticized loan | Older than 36 months | 12 · 24 · 36 months |
| Cooling-off after moving from lending | 12 months | 6 · 12 · 24 months |
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 bank’s data · “target” = the design goal, measured in the live solution · “estimated” = our estimate of where a full-file review’s hours go · “typical” = published figures (bank 10-K and 10-Q loan review coverage targets of 35–80% a year; Abrigo, reviewers oversee $500M–$1B each). Context: Interagency Guidance on Credit Risk Review Systems (SR 20-13, 2020). People, banks and borrowers named on this page are fictional — characters and sample data in the working solution.
Software for a bank’s loan review team — the independent check on the grades lenders give their own loans. Credit Risk Review re-grades every commercial loan from its file each quarter, shows every loan whose grade of record the file no longer supports, and carries each one through the reviewer’s decision, the challenge to the lender and the board report.
The agents assemble the credit file, spread the statements on the credit agreement’s own definitions, recompute every covenant and score each factor of your rating scale — repayment capacity, leverage, liquidity, concentration, collateral, management, guarantor and payment performance. The suggested grade comes with every reason cited to a page and why it is not one notch better or worse.
No. The agents suggest a grade; a loan review officer sets the review grade. The grade of record changes only when the challenge with the lender is settled, and every decision is recorded with who, when and why.
It reads every commercial loan every quarter instead of a yearly sample, so reviewers spend their hours on the loans where the file and the grade disagree. Loans over a set size, insider loans and the watch list can still always go to a person, even when the agents agree.
The financial spreader recomputes EBITDA on the agreement’s definition and accepts an add-back that needs the bank’s approval only if the approval is in the file. The covenant tester compares every certificate with the recomputed value and reports a difference even when both pass.
The review prioritiser never assigns a reviewer to a loan they approved or managed, checked against the approval history of every loan, and keeps reviewers off their former region for a cooling-off period you set.
The lender answers the challenge within the response clock — 10 business days by default. If they still disagree, the review write-up, the lender’s answer and every cited page go to the Chief Credit Officer, who decides.
The Agentic Solution Engine builds and deploys it from your requirements — your credit policy and rating scale, your spreading template, your challenge format and a sample of credit files — and it goes live once every quality gate has passed. We will walk you through it on your own loans first.
Grade drift seen within the quarter, before the examiners pull their sample.
1-min demoCECL Allowance ManagementEvery qualitative adjustment arrives with its support already written.
1-min demoCredit Concentration Risk MonitoringLimits checked daily, not weeks after the quarter closes.
1-min demoWe’ll run Credit Risk Review on a sample of your own commercial credit files.