An internal credit rating for every issuer, cited to its page and signed by your credit committee
Every rating cited to its page, and every two-notch split explained before committee.
See one case, screen by screen ↓Internal Credit Rating is software for the internal credit rating a buy-side credit team keeps on every issuer it holds. Seven AI agents spread the filings with every value cited, score them under the firm's approved credit rating methodology, place the issuer among peers, flag any rating two or more notches from the agencies with a drafted reason, and answer SEC, NAIC and Central Bank requests from the file. The analyst proposes, an independent reviewer challenges and the committee chair signs.
AI in asset management: a buyer's guide to agentic AI for investment managers →
Updated
An internal credit rating is the buy-side credit team's own documented view of an issuer, and for many portfolios the rules now require one. Since October 2016 a US money market fund adviser can no longer rest eligibility on agency ratings: it documents a minimal credit risk determination for every name against four factors. EU fund managers may not rely solely or mechanistically on ratings, and an EU money market fund reviews every internal credit assessment at least once a year. Since January 1, 2026 the NAIC's Securities Valuation Office may challenge a filing-exempt designation three notches or more from its own view. Yet the rating file is still a workbook, a memo and an email chain.
The risk sits in what nobody wrote down. The spread is re-keyed and the source page behind a number is lost when the analyst moves on. The methodology's weights live in a PDF. A notch off the scorecard is explained in one sentence, if at all, and a two-notch split from the agencies is argued in the room rather than recorded. The independent challenge happens by email, the minutes are typed afterwards, and when the SEC, the NAIC or the Central Bank of Ireland asks for the file, it is rebuilt for weeks. In 2019 the SEC staff found money market credit files missing required factors.
Estimated split for one annual review of a high-yield corporate issuer, by hand and with the solution.
Seven agents spread, score, place and check every issuer; an analyst proposes, a reviewer challenges, the chair signs.
Orrin Freight Lines, a US truckload and intermodal carrier in Columbus, Ohio, is due its annual review. Quillfield Capital holds $412.6M of its bonds in six portfolios, two of them for insurance clients. The Credit Committee meets at 2:00 p.m. Here is the day, screen by screen, in the working solution.
Maya covers 84 issuers in Industrials & Transportation. Her home screen shows 9 reviews due by Oct 30, 1 split needing her reason and 6 values to confirm. Overnight the agents read 23 new filings, 4 not in English, opened an event review for Sauveterre Transports on a French profit warning, and checked 174 money market credit files for the SEC. Every line is cited, and under it she can ask the agents, or tell them what to do — “Why is Orrin three notches below?”
“Orrin Freight Lines: 3 notches below the agencies. You proposed BB−; agencies sit at BBB−. The Divergence agent drafted a cited reason.”
The agents state their plan, then work through it: the Intake agent checks the 10-K FY2025, the 10-Q for Q2 2026 and Exhibit 10.4; the Spreading agent reports 96 values, 94 above 0.90 confidence, totals tied out; the Scorecard agent scores 8 factors to a weighted 11.60, BB; the Peer agent places Orrin among 9 transportation peers. The Divergence agent finds the gap and the Challenge agent drafts 2 questions for the reviewer.
“Hand-off to Maya Lindqvist. Nothing changed in the rating; the file waits for a person.”
AAA to D runs down the center of the file. Quillfield's Corporate Methodology v4.2 scores Orrin BB. Maya notches down once for customer concentration — the largest customer is 38% of revenue — and leaves an owned-fleet notch unapplied because it is already in the liquidity score. All three agencies sit at BBB−, so her BB− is three notches below, on bonds two insurance clients hold.
Divergence agent: “A reason is required before review. I drafted one from the evidence.”
Clicking the citation beside the adjustment opens the 10-K, Item 1A, risk factors: “approximately 38% of our operating revenues in 2025 (36% in 2024)”. The panel says who read it, the confidence (0.98) and where it is stored in the rating file. Two years from now, the page is still one click away.
The memo, now at version 3, carries the recommendation, rating drivers, the adjustment, the divergence section and the peer position. Every sentence links to its page — 18 citations, each checked against its source. A four-factor section, the Rule 2a-7 factors, is written for every file, so a money market eligibility question is answered from the same memo.
Maya edits one sentence of the drafted reason and approves it, and the file goes to Tom in Credit Risk Review in London. The Challenge agent has drafted two questions; Tom sends one: “Is 38% durable? The agreement runs to 2029.” Maya answers from the file — the customer can end the agreement on 12 months' notice and cut volumes 22% in 2023. Tom accepts the adjustment with the reason code “Evidence sufficient”.
The Challenge agent drafts the questions. It never sends a challenge, and never accepts or rejects.
The agenda holds 4 rating changes, 2 new issuers and 12 affirmations, with quorum 4 of 6 voting members across New York and London. Joanna Kerr is recused from Quinnell Digital Infrastructure because her team arranged the private placement. The File agent drafts the minutes as the meeting runs.
“Committee pack ready: memo v3, scorecard, 14 source passages and the challenge record. Votes in: 5 of 6 — 4 for BB−, 1 abstention.”
Linda did not propose or review this file, so she may sign it. She picks the reason code “Rating change: concentration” and attests that she has read the file and the challenge record, and signs BB− as chair of the Credit Committee, effective October 14, 2026.
“I have read the file and the challenge record, and I sign this rating as chair of the Credit Committee.”
The signed rating goes to the risk system and portfolio holdings at 2:41 p.m. and to the order management watch list at 2:42 p.m., after one automatic retry. Marcus Hale is told the Brightwater investment-grade account is now below its floor. A client notice to the two insurance clients is drafted and waits for Rafael Moreno, the CRO.
The File agent never sends anything outside the firm; the CRO approves every client notice.
SEC document request item 14 asks for the credit file behind every money market fund holding. The File agent maps it to 174 credit files — 148 prime, 26 government and repo — and checks each against Rule 2a-7: 171 carry all four factors and a review inside 12 months. The 3 gaps are listed first with their owners, due Tue Oct 20, and the pack cannot be approved while a file is incomplete.
“Every request mapped to the rating files, checked against its rule, gaps first · the Chief Compliance Officer approves every response.”
The same rating file, seen by five of the people who carry it — what their review cycle looked like, and what it looks like now.
Seven agents spread, score, place and check every issuer; an analyst proposes, a reviewer challenges, the chair signs.
Watches filings, agency actions, news and covenant events; opens annual and event reviews with a due date and cites the event.
Spreads 10-Ks, 10-Qs and annual reports into the house template in the issuer's own language, with lease, pension and one-off adjustments; ties every total out and cites each value to its page.
Applies the approved credit rating methodology version and sector table; scores each factor and computes the weighted score and the implied rating.
Picks peers by sector, size and rating band, places the issuer on leverage, coverage, margin and concentration, and flags where it is the outlier.
Compares each proposed rating with the three agencies and their median; at two notches or more it stops the file and drafts a cited reason, and at three notches it checks which insurance clients hold the bonds and drafts a client notice.
Reads the file as a skeptical reviewer would — double counting against the methodology, stale inputs, peer inconsistency, reasons that do not match their source — and drafts questions for the independent reviewer.
Drafts the rationale memo with the Rule 2a-7 factors and the committee minutes; maps SEC, NAIC and Central Bank requests to the files, checks each against its rule, lists the gaps first and builds the pack.
Maker, checker, signature, methodology. The agents propose; a named person decides.
An ask bar sits on every screen. Answers cite their source and open the screen; asked to act — remind the analysts, re-check a name when a feed lands — the agents show their plan and wait for a person to approve or decline.
Why is Orrin Freight Lines three notches below the agencies?
The scorecard gives BB (weighted 11.60 under Methodology v4.2), and Maya took one notch for customer concentration: the largest customer is 38% of revenue. The agencies, all BBB−, weight the contract that runs to 2029; it can be ended on 12 months' notice, and that customer cut volumes 22% in 2023.
Which splits still need a reason?
9 of 47 splits of two notches or more have no approved reason yet. Orrin Freight Lines is first: it is on today's committee agenda.
Are the money market credit files ready for the SEC?
171 of 174 files carry all four Rule 2a-7 factors and a review inside 12 months. Three gaps remain: Ashgrove Bancorp (liquidity update over 12 months), a repo counterparty review, and a missing stress case for Larchmont Foods. The response is due Fri Oct 23.
What would methodology v4.3 change?
Re-scored under v4.3, 9 of 61 transportation issuers move: 7 down one notch and 2 up one notch; 52 stay. Orrin would score BB− directly, because concentration becomes a scorecard factor instead of an adjustment. Validation has 3 of 5 tests passed.
13 screens from the working solution, on its sample data. Pick one to see it large.
The thresholds every internal credit rating is checked against are settings. Each change is made by the administrator, approved by the CRO and logged; the current value stays in force until then.
| Setting | Default | Choose from |
|---|---|---|
| Reason required when a rating is this far from the agency median | 2 notches | The Divergence agent stops the file and drafts a cited reason |
| Insurance holder alert (NAIC challenge level) | 3 notches | Client notice drafted for the CRO |
| Annual review cycle | 12 months | Money market names: also on any credible event |
| Event review due | 10 business days | From the event date |
| Independent review SLA | 3 business days | A change to 2 business days waits for the CRO |
| Escalation when late | 5 days → Head of Credit Research · 15 days → CRO | Set by the administrator, approved by the CRO |
| Adjustment limits | ±2 per adjustment · ±3 in total | From the methodology (§5.1) |
| Value confidence needing a person | below 0.90 | Values below it go to the analyst to confirm or override |
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 · “typical” = published figures (SEC staff Risk Alert on money market funds, November 2019; NAIC issue brief, Leveraging Credit Ratings for Regulatory Use, March 2026). Regulatory context: SEC Rule 2a-7 amendments removing references to agency ratings, compliance October 14, 2016; EU Money Market Fund Regulation 2017/1131, Articles 19–21; CRA Regulation Article 5a. People, firms, issuers, funds and insurance clients named on this page are fictional — characters and sample data in the working solution; the three rating agencies appear there as Agency A, B and C.
Internal Credit Rating is software for the internal credit rating a buy-side credit team keeps on every issuer it holds. Seven AI agents spread the filings with every value cited, score them under the firm's approved credit rating methodology, place the issuer among peers, flag any rating two or more notches from the agencies with a drafted reason, and answer SEC, NAIC and Central Bank requests from the file. The analyst proposes, an independent reviewer challenges and the committee chair signs.
A review opens on the 12-month cycle or on an event such as a profit warning or an agency action. The filings are spread, the methodology's scorecard gives an implied rating, the analyst notches up or down for what the scorecard misses, an independent reviewer from Credit Risk Review challenges, and the credit rating committee votes and its chair signs. In this solution the agents do the reading and drafting at each step, and every decision is made and signed by a person, with a reason code.
Since the SEC removed agency-rating references from Rule 2a-7 (compliance October 14, 2016), a money market fund's adviser documents its own determination that a security presents minimal credit risks, weighing the issuer's financial condition, sources of liquidity, ability to react to future market-wide and issuer-specific events, and the strength of its industry. The File agent writes and cites all four in every memo, and checks every money market credit file for the four factors and a review inside 12 months.
The Divergence agent stops the file and drafts a reason from the evidence, with its sources; the analyst approves, edits or rejects it, and a reason that cites nothing is refused. The file cannot go to independent review without one. At three notches it checks which insurance clients hold the bonds and drafts a client notice for the CRO.
Yes. From January 1, 2026 the NAIC SVO may challenge a filing-exempt designation three notches or more from its own view, and an information request gives 45 days, plus a possible 45-day extension. The solution shows the clock, maps the request to the rating file, checks the rationale against the standard of a public rating report and assembles the pack for the CCO to approve. In the demo, the internal rating is BBB, two notches below the notes' private letter rating of A−, and the file explains why.
Yes. Your internal credit rating model is your approved methodology versions and sector tables; the Scorecard agent applies only those and cannot change a weight or notch. A proposed version is re-run on the book first — in the demo, v4.3 moves 9 of 61 transportation issuers by one notch — validated by Model Risk and approved by the CRO, never by its proposer.
Yes. The agents spread, score, compare and draft; they never change a rating, sign, send a challenge or send anything outside the firm. The analyst, the independent reviewer and the chair are three different people on every file, and every step is on a hash-chained audit trail.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your rating methodology, rating scale, coverage book and a sample of past rating files — and it goes live once every quality gate has passed. A five-step setup connects the systems, imports the issuers, maps the rating scale, chooses the methodology and assigns analysts and reviewers, then goes to the CRO for approval.
Every credit question answered in minutes, each sentence pinned to its tranche and page.
1-min demoBond Covenant AnalysisEvery basket in dollars and every leakage route found, before books close.
1-min demoProspectus ReviewEvery deal that fits a fund briefed and decided before books close, every line cited.
1-min demoWe'll run Internal Credit Rating on a sample of your own issuers, your methodology and last cycle's rating files.