SQL · Find the facts
The question
Which required fields are missing or inconsistent for this application?
Check structured evidence.
Explore SQLLending
Review loan applications
with their source documents.
Find missing documents, conflicting borrower details and the applicable lending policy. Keep the source and date visible for the underwriter.
Application review
Check the application documents and identify what is missing.
Illustrative use case01 / Use case guide
10 practical use cases.
Compare the data models behind everyday industry tasks, from finding records to following connections.
SQL Filter and calculate Vector Match similar meaning Graph Follow relationships Full-text Find words and phrases
Suggested model combination
Scroll across to compare all four models| Use case | How the data models help | SQL | Vector | Graph | Full-text | Source |
|---|---|---|---|---|---|---|
| Loan application completeness review Underwriters need to identify missing or inconsistent application evidence. | SQL checks required application fields, while Graph links each borrower claim to its submitted supporting documents. | SQL included | Vector not needed for this example | Graph included | Full-text not needed for this example | |
| Self employed income evidence review Underwriters need to reconcile business income across tax returns and financial statements. | SQL compares income periods, Vector retrieves relevant income guidance, and Full-text locates supporting tax schedules for underwriter review. | SQL included | Vector included | Graph not needed for this example | Full-text included | |
| Property appraisal evidence review Collateral reviewers need to investigate inconsistent property facts and appraisal commentary. | SQL compares recorded property facts, Vector finds relevant review guidance, and Full-text retrieves the appraiser's supporting explanations. | SQL included | Vector included | Graph not needed for this example | Full-text included | |
| Business loan covenant monitoring Credit teams need to check reported financials against agreed loan conditions. | SQL calculates agreed reporting measures, while Full-text retrieves the covenant clauses for a credit officer's review. | SQL included | Vector not needed for this example | Graph not needed for this example | Full-text included | |
| Construction loan draw review Loan administrators need inspections, budgets and lien records before reviewing a draw request. | SQL compares requested draws with budgets, Graph links inspections and lien records, and Full-text retrieves unresolved inspection findings. | SQL included | Vector not needed for this example | Graph included | Full-text included | |
| Mortgage closing document comparison Closing teams need to explain changes between loan estimates and final disclosures. | SQL compares recorded rates and fees, while Full-text retrieves explanations in the corresponding document versions. | SQL included | Vector not needed for this example | Graph not needed for this example | Full-text included | |
| Mortgage escrow discrepancy investigation Servicers need to explain mismatches among escrow collections, tax bills and insurance payments. | SQL reconciles escrow amounts and dates, while Graph links the loan to tax bills, insurance invoices and disbursements. | SQL included | Vector not needed for this example | Graph included | Full-text not needed for this example | |
| Borrower hardship package review Servicing specialists need complete, current evidence before assessing a request for mortgage assistance. | SQL checks document dates and completeness, Vector retrieves relevant assistance guidance, and Full-text finds supporting hardship statements for specialist review. | SQL included | Vector included | Graph not needed for this example | Full-text included | |
| Loan credit reporting dispute review Servicers need to investigate a borrower's disputed balance or repayment history. | SQL compares ledger and reported values, Graph links reporting submissions to the loan, and Full-text retrieves the borrower's evidence. | SQL included | Vector not needed for this example | Graph included | Full-text included | |
| Loan payoff quote verification Servicers need to explain why a dated payoff quote differs from the displayed balance. | SQL reconciles principal, accrued interest and recorded fees, while Full-text retrieves relevant payoff terms for servicing review. | SQL included | Vector not needed for this example | Graph not needed for this example | Full-text included |
Use the query interfaces you need, then connect their results through shared record IDs in your application.
Industry sources 10
These sources describe real industry workflows. The model combinations above are proposed OriginChainDB designs, not customer deployment claims.
- Fannie Mae — Contents of the Application Package (opens in a new tab)
- Fannie Mae — Underwriting Factors and Documentation for a Self-Employed Borrower (opens in a new tab)
- Fannie Mae — Review of the Appraisal Report (opens in a new tab)
- Office of the Comptroller of the Currency — Comptroller's Handbook: Commercial Loans (opens in a new tab)
- Office of the Comptroller of the Currency — Comptroller's Handbook: Commercial Real Estate Lending (opens in a new tab)
- Consumer Financial Protection Bureau — Review documents before closing (opens in a new tab)
- Consumer Financial Protection Bureau — Problems with escrow or impound accounts (opens in a new tab)
- Fannie Mae — Receiving a Borrower Response Package (opens in a new tab)
- Consumer Financial Protection Bureau — How to dispute an error on a credit report (opens in a new tab)
- Consumer Financial Protection Bureau — Payoff amounts and current balances (opens in a new tab)
02 / A closer look
Checking an incomplete business loan application
A small-business applicant submits documents across several channels. The reviewer must reconcile consented account information, business identifiers and the current lending policy before requesting more evidence.
What makes it difficult
- 01
Different submissions can describe the same business differently.
- 02
Missing or outdated evidence creates repeated follow-ups.
- 03
A reviewer needs the applicable policy version, not the nearest-sounding passage.
Suggested approach
Check completeness before eligibility
Use structured checks for required borrower facts. Search approved policy documents for supporting clauses, restricted to the correct loan product and policy version.
03 / Queries
What each data model does
Select a model to see the question it answers.
Vector · Match the meaning
The question
Which approved policy passages address irregular business income?
Find relevant policy explanations.
Explore Vector Graph · Follow the connections
The question
Which documents and consent records belong to this borrower and business?
Connect the application and its evidence.
Explore Graph Full-text · Find the exact wording
The question
Where does this application number or policy clause appear?
Match exact identifiers and terms.
Explore Full-textUse shared record IDs to combine results from the different query interfaces in your application.
04 / Implementation
How to build it
- STEP 01
Choose a loan product
Use de-identified applications and an underwriter-approved checklist. Define permitted data uses, retention and reviewer access.
- STEP 02
Record document provenance
Keep application, business, document and consent IDs. Store dates, extraction confidence and policy versions. Update indexes when documents change or consent is withdrawn.
- STEP 03
Build the review brief
Select the application and policy version with SQL. Retrieve relevant passages and linked documents. Mark missing or conflicting facts in the application.
- STEP 04
Review with underwriters
Compare completed applications with expert annotations. Test outdated policies, withdrawn consent and restricted documents before a supervised pilot.
05 / Testing
What to measure
Compare these results with your current process before rollout.
Evidence completeness
Compare the brief against an underwriter-authored checklist for each application.
Policy accuracy
Check whether retrieved passages belong to the correct product, jurisdiction and effective version.
Follow-up effort
Measure repeated requests caused by missing, inconsistent or hard-to-find evidence.
06 / Sources
Examples from the industry
Published projects and reference material relevant to this use case.
FinBox & IIFL Finance
Cash-flow and GST-based business lending
FinBox reports working with IIFL Finance on an MSME lending product using cash-flow and GST data through the Account Aggregator ecosystem. This documents a real lending workflow, not an OriginChainDB implementation.
Read the sourceSahamati & CRIF
Consent to Credit: Account Aggregator impact report
The report describes Account Aggregator use in underwriting and monitoring, with case studies spanning home loans, small-business credit and new-to-credit borrowers. Its findings concern the AA ecosystem, not database performance.
Read the sourceThese are industry references, not OriginChainDB customer stories. The implementation above is a proposed example.
Build with OriginChainDB