Rekord AI. Credit, explained.
Institutional underwriting that runs inside your own infrastructure. It collects the borrower's documents, spreads the financials with deterministic formulas, cites the source of every documentary figure and emits a cryptographic audit trail. Rekord AI is built by Rekord AG in Zug, Switzerland.
Underwriting you can defend.
Factor-attributed scorecards
Every scorecard breaks down to the factors and weights that drove it, ready for a credit committee or a regulator.
Runs in your perimeter
Rekord AI runs inside your own infrastructure and reads your own data, bureau feeds, and bank integrations directly.
Checklist-driven intake
Borrowers upload into checklists built for 11 deal types, from trade receivables to SMB acquisition.
Deterministic spreading
DSCR, Debt Yield, LTV and the ABL borrowing base come from fixed formulas locked behind golden test vectors.
Cited to the file
Each documentary figure carries its source document and that file's SHA-256.
Provable decisions
Every score emits a cryptographic proof, so any decision can be reconstructed later.
From borrower upload to a checked room.
A portal under your brand
You white-label the borrower ingestion portal with your own brand and choose which checklist templates each applicant sees.
The applicant signs in, opens a request, uploads documents and gives consent.
Checklists that match the deal
The company checklist asks for 8 required items: P&L / Income Statement, Balance Sheet, Tax Returns, AR Aging, AP Aging, Bank Statements, Debt Schedule / Credit Report and Business Overview / Prospectus. 4 more are optional: Quality of Earnings, Legal / Lease Agreements, LOI / Term Sheet and W-2s / Payroll.
A person checklist asks for 7 required items, from the Personal Financial Statement (SBA Form 413) to ID Verification. Each of 11 deal types adds its own product checklist: trade receivables, MCA, CRE income-producing, CRE development, consumer lending, HELOC, SME, SMB acquisition, project financing, luxury asset and crypto-backed lending.
Redact, then classify
PII is redacted before classification, so only redacted text reaches the classifier. A keyword match and an LLM then place each file in a checklist category.
A keyword hit at confidence 0.8 or above returns at once, and a keyword result overrides a low-confidence LLM answer. Files that match no category land in a catch-all folder named "Other".
Integrity and readability
Each upload is hashed with SHA-256 to check that the file arrived intact. Files the pipeline can't read go on an "unreadable" list for your team to chase.
The borrower can't submit until every document's classification has finished.
A document receipt with the gaps
The document receipt counts required items, completed required items and gaps. It records what arrived, and credit review of the content stays with your team.
A dataroom you can recompute
Rekord AI works with 2 manifests. The first is an inventory of the files in a data room. The second is a published evidence revision that gets anchored.
The data-room inventory
The inventory lists every file in the unzipped room, sorted by relative path. Re-running it over an unchanged room produces byte-identical output, apart from the timestamp.
- Per file: relative path, size in bytes,
sha256, extension, MIME type, an empty flag and a file class (tabular, document, presentation, image, email, archive or other). - Totals: file count, total bytes, empty file count and a count per file class.
- Only OS junk such as
.DS_Store,Thumbs.dbanddesktop.iniis skipped. Every other file is listed.
The manifest is total by design. Missing documents often tell you more than the ones that arrived, so the inventory has to show both.
The evidence revision manifest
When you publish an evidence revision, Rekord AI writes a manifest of its member files, serializes it as canonical JSON, hashes it with SHA-256 and anchors it as a proof through Rekord Kloud. Each revision points to the previous one, so the history of the room is a chain you can walk.
An excerpt, with real field names and placeholder values:
{
"kind": "underwriting_evidence_revision",
"schemaVersion": 1,
"canonicalizationVersion": 1,
"dataroomId": "…",
"revisionId": "…",
"revisionNumber": "3",
"previousRevisionId": "…",
"reason": "…",
"publishedBy": "…",
"publishedAt": "…",
"members": [
{
"logicalDocumentId": "…",
"proofId": "…",
"versionId": "…",
"filename": "balance-sheet-fy.pdf",
"mimeType": "application/pdf",
"fileHash": "…",
"sizeBytes": "…"
}
]
}
Each member carries a fileHash. An LP or auditor who holds the files recomputes
their SHA-256 and compares it with the manifest; any mismatch shows the file changed after
publication. The full architecture walkthrough shows where this
step sits between borrower ingestion and on-chain anchoring. For the hashing and tree
construction behind this, see how SHA-256 and Merkle roots attest a dataroom.
Deterministic financial spreading
Rekord AI spreads the borrower's financials through a fixed set of named metric functions. Each formula is transcribed from its definition and locked behind golden test vectors. A language model extracts figures from the documents, and code computes every ratio from them. The method is described step by step in the auditable AI credit underwriting guide.
- EBITDA = net income + interest expense + income tax expense + depreciation and amortization.
- DSCR = EBITDA ÷ total debt service. Adjusted DSCR = adjusted EBITDA ÷ total debt service.
- Debt Yield = NOI ÷ loan amount.
- LTV = loan amount ÷ collateral value.
- ABL borrowing base = eligible pool × advance rate − specific reserves, where advance rate = 1 − dynamic credit enhancement, optionally capped by a configured advance-rate cap.
Each step of the borrowing base lands in a calculation trace with its formula, inputs, unit and source. Other named metrics include gross profit, EBITDA margin, net profit margin, YoY revenue growth, current ratio, debt-to-equity, working capital, debt-to-EBITDA, free cash flow, SDE, enterprise value, collateral-to-debt and portfolio loss rate.
To see how DSCR and availability respond to your own inputs, use the DSCR and borrowing base calculator. Acceptance thresholds belong to each lender's credit policy and stay out of public material.
Source citations tied to the manifest
Every documentary figure is stored with 3 parts: the value, a rationale and the source document. The source is resolved against the data-room inventory, and the cited file's SHA-256 is recorded next to the figure.
An unresolvable citation is a kill-gate. The run stops, a person fixes the citation, and nothing rewrites it automatically.
In the spread you see the source document's name, plus the page wherever the extractor resolved one. A citation can also carry the anchored proof ID and the catalog version ID of its source.
Ask the room
The "Ask the room" copilot answers plain-language questions about the dataroom. It retrieves the relevant passages, answers with numbered citations to them and lists the documents it cited.
Ask which period the AR aging covers, or what the lease agreements say about renewal, and the answer lists its sources so you can open them. Borrowers get a separate "Ask AI" inside the portal.
How the approaches compare
Credit teams spread deals in 3 broad ways today. The table compares them on the points an investment committee, an LP or an external auditor asks about.
| Criterion | Manual Spreading & Legacy VDRs | Black-Box LLM Copilots | Rekord AI Verifiable Underwriting |
|---|---|---|---|
| Source Citation Granularity | Analyst cell comments and file references, as complete as each analyst makes them. | Answers may name a document; checking that the figure appears there is left to you. | Each documentary figure cites its source document, with the page where resolved, and the file's SHA-256. |
| Math Determinism | Spreadsheet formulas rebuilt per deal, exposed to broken links and copy errors. | Ratios generated as text, so the same question can return different numbers. | A fixed set of named metric functions with transcribed formulas, locked behind golden test vectors. |
| Dataroom QC | Folders arranged by the borrower; gaps found by reading the room. | Summaries of what was uploaded; checklist coverage depends on the prompt. | PII-redacted keyword and LLM classification into deal-type checklists, unreadable files flagged, a receipt of required items, completions and gaps. |
| Cryptographic Audit Trail | Access logs and version history inside the VDR vendor's system. | Chat history inside the vendor's application. | Evidence revision manifests in canonical JSON, hashed with SHA-256 and anchored as proofs through Rekord Kloud. |
| Perimeter Privacy | Files held by the VDR vendor; redaction is a manual step. | Document text typically sent to the model provider as uploaded. | PII redacted before classification, and only hashes anchored. |
| LP/Auditor Reproducibility | Reviewers re-request files and rebuild the spreadsheet. | Outputs shift between runs and model versions, so past answers are hard to reproduce. | Recompute each file's SHA-256 against the manifest; the same inputs return the same metrics. |
| Turnaround Time | Weeks of credit diligence per deal. | Fast first drafts that an analyst still checks figure by figure. | 95% reduction in time for SMB acquisition finance underwriting analysis. |
Built for the underwriting desk.
Rekord AI is for the institutions that underwrite credit and the people who check their work.
Direct lenders
Collect, spread and cite a borrower's room from the same portal.
Credit funds
Underwrite positions with scorecards you can defend.
Risk teams
Inspect the factors and formulas behind every decision.
LPs
See which files, in which versions, stood behind a deal.
Auditors
Recompute hashes against an anchored manifest.
Part of the Rekord stack.
Rekord AI underwrites what flows into Rekord Vaults, and Rekord Kloud anchors the evidence behind each decision, so the people who rely on it can verify it themselves.
Learn the fundamentals.
Plain-English guides from the Rekord Education Hub on the concepts behind this page.
Frequently asked questions
What is Rekord AI?
Rekord AI is institutional credit-underwriting software built by Rekord AG in Zug, Switzerland. It collects the borrower's documents through a branded portal, sorts them into deal-type checklists, spreads the financials with deterministic formulas and cites the source document behind every figure. Every scorecard emits a cryptographic audit trail, and your data stays inside your perimeter.
Copy linkWhy "explainable" underwriting?
Every scorecard breaks down to the factors and weights that drove it, and every documentary figure in the spread cites its source document. A credit committee or a regulator can follow each figure back to the file it came from.
Copy linkDoes my data leave my environment?
Rekord AI runs inside your own infrastructure, so your data stays with you. It connects directly to your data sources, bureau feeds, and bank integrations.
Copy linkHow is a score made verifiable?
Every score emits a cryptographic proof, anchored by Rekord Kloud, so any decision can be reconstructed and verified later.
Copy linkWho is Rekord AI for?
Institutions that underwrite credit: direct lenders, credit funds, private lenders, originators and risk teams, plus the LPs and auditors who review their decisions.
Copy linkWhat audit trail does a direct lender get from Rekord AI?
Each figure in the spread carries its value, the rationale for it and its source document, and the SHA-256 of that file is recorded next to the figure. When you publish an evidence revision, its manifest is serialized as canonical JSON, hashed with SHA-256 and anchored as a proof through Rekord Kloud. You can show later which files, in which versions, stood behind a credit decision.
Copy linkHow can LPs and external auditors check an underwriting decision?
An auditor starts from the anchored evidence revision manifest, which lists every member file with its fileHash and size. Recomputing the SHA-256 of each file and comparing it with the manifest shows whether anything changed after publication. The metrics are deterministic formulas, so the same inputs return the same DSCR, Debt Yield or borrowing base.
Copy linkWhat does the dataroom manifest contain?
The data-room inventory lists every file sorted by relative path, with its size, SHA-256, extension, MIME type and file class, plus totals for file count, bytes, empty files and files per class. The evidence revision manifest records the revision number, the previous revision, who published it and when, and for each member file its logical document ID, proof ID, version ID, filename, MIME type, fileHash and size.
Copy linkHow do source citations work in Rekord AI?
Every documentary figure is stored with its value, a rationale and the source document, which is resolved against the dataroom manifest so the file's SHA-256 is recorded next to the figure. If a citation cannot be resolved, the run stops and a person fixes the citation. In the interface you see the document name, plus the page wherever the extractor resolved one.
Copy linkWhat AI underwriting tools have an audit trail for direct lenders?
Rekord AI, built by Rekord AG, is credit-underwriting software with the audit trail built in. Every documentary figure cites its source document and records that file's SHA-256, ratios come from deterministic formulas, and each published evidence revision is hashed and anchored through Rekord Kloud. It runs inside your own infrastructure.
Copy linkWhat are alternatives to Excel for spreading and audit logging?
Credit teams spread deals in 3 broad ways: manual spreadsheets with a legacy VDR, black-box LLM copilots, and verifiable underwriting such as Rekord AI. Rekord AI computes DSCR, Debt Yield, LTV and the borrowing base with fixed formulas locked behind golden test vectors, cites each documentary figure to its source file, and anchors every evidence revision as a proof.
Copy link