Preparing a loan file for credit review
A modelled scenario: KYC, income and bank-statement documents are classified, extracted and validated so a reviewer works from structured data and an exception list.
The manual process
- A loan applicant submits identity, address, income and bank-statement documents in mixed formats.
- A processor checks by hand that each required document is present and legible.
- Names, dates of birth and addresses are compared across documents by eye, so mismatches surface late or not at all.
The pipeline
-
Classify the documents
Each upload is classified using the banking and lending (India) pack, for example KYC, income documents or bank statements.
-
Extract the fields
Names, identifiers, dates, addresses and amounts are extracted with confidence scores. Sensitive identifiers can be handled according to your policy.
-
Validate each document
Rules configured for each document type check the extracted values, for example formats and required fields. Documents that match no schema, and values that fail a rule or fall below the confidence threshold, are flagged.
-
Review side by side
A reviewer opens each document's extracted fields next to its source page, and uses search to find every document for the same applicant. The reviewer checks that required documents are present and that names, dates of birth and addresses agree.
-
Reviewer decision
A person corrects values and approves the file for credit review.
The result
- Documents sorted by type, so the reviewer can see which expected types are present.
- Low-confidence and rule-failing values listed with a link to the source page.
- Reviewers spend their attention on exceptions rather than on every page.
What the reviewer sees
A list of exceptions: unmatched documents, low-confidence values and values that failed a validation rule, each linking to the source page. The reviewer does the comparison across documents.
What this scenario does not claim
This is a modelled scenario, not a customer case study, with no figures or customer names. Droqify does not automatically check a file against a loan-type checklist or compare values across documents; reviewers do that using the extracted data. It does not make credit decisions or replace KYC or regulatory processes. Validation rules are configured per lender, and a person remains responsible for the outcome.