AI & Analytics
A cost-effective way to turn complex loan files into clear, reviewable decision packages.
ZYGY Loan Approval Intelligence brings application documents, data checks, credit policies and review workflows into one process. Lenders spend less effort re-keying files and chasing gaps, and credit officers receive a reasoned recommendation instead of a score on its own.
The lending workflow
A single application may include forms, identity documents, payslips, bank statements, employment confirmation, existing loan information, financial statements, collateral documents and credit reports.
Officers retype the same facts from forms, payslips and statements into the credit file.
A file can move forward before anyone notices an expired, duplicated or absent document.
Names, income and commitments often disagree across the application, statements and credit report.
Review starts late because the package still has to be gathered, checked and explained.
Why ZYGY
The value is a lower-cost review process: one workflow instead of separate document handling, scoring and bureau checks. Conventional processing can return a score and leave the explanation to the credit officer. ZYGY adds the reasoning — what it found, which documents support it, and where the file still conflicts — and can bring third-party credit bureau results into that same review.
The package can include a risk indicator and a written explanation: which facts were used, which documents disagree, and what the officer should check next.
Each finding points back to the application, payslip, statement or bureau item that produced it, so the recommendation can be traced.
The workflow can request and reconcile data from third-party credit bureau agencies, then compare facilities and monthly obligations with what the applicant declared.
Use cases
The same decision package applies across products. Credit policy, documents and approval limits stay with the institution.
Payslips, bank statements and declared commitments are checked before a credit officer reviews the recommendation.
Application, income and collateral documents are reconciled against the institution’s loan-to-value and document rules.
Identity, income, existing facilities and property documents are assembled into one reviewable file.
Financial statements are tested for consistency, then set beside bureau data and the institution’s credit policy.
Decision intelligence
Documents move through generative, statistical and predictive checks, then through the institution's policy and workflow. Generative intelligence supplies the reasoning: summaries, conflicts and questions. Predictive checks can add a risk indicator. Neither one replaces the credit officer. Bureau data can enter the same path. The institution keeps its credit policy, approval authority and final lending decision.
Documents
Application files and supporting evidence enter one process.
Generative intelligence
Reads unstructured documents and prepares summaries, explanations and reports.
Statistical intelligence
Tests financial information and supports affordability, variance and consistency checks.
Predictive intelligence
Can assess likely risk using historical patterns and application features.
Policy and workflow
Applies the financial institution’s configured rules and review process.
Explainable decision
Presents findings, evidence, exceptions and a recommendation for human review.
Application workflow
The workflow prepares a decision package. It does not approve or decline the loan.
Submit application
Classify documents
Extract information
Validate and reconcile
Apply credit rules
Prepare decision recommendation
Review outcomes
Cross-document checks
ZYGY builds one applicant record from the file, then checks whether the evidence is complete and consistent.
Application
Payslips
Bank statements
Credit information
Unified applicant record
Validation and reconciliation
Name, identification number, date of birth and address.
Income amount, payment frequency and salary-credit consistency.
Existing facilities and monthly obligations.
Revenue, expenses, assets, liabilities and mathematical consistency.
Missing, expired, duplicated, illegible or conflicting documents.
Credit and affordability
Approved policies can be configured into the workflow. The solution supports consistent application of those policies. It does not replace the institution’s credit policy or approval authority.
Decision package
The figures below are a fictional sample, labelled as an illustrative example.
Illustrative example
Refer for credit review
Illustrative example. Bars share one scale.
Five findings in this sample file.
Combined commitments and the proposed instalment are RM4,600 a month against verified income of RM8,200. The sample marks affordability as borderline.
This package is a recommendation for human review. It is not a lending decision.
Human review
A credit officer can review evidence, request information, record conditions, make or change the final decision, and document any override reason.
Illustrative example
Name differs between the application and a bank statement.
Next action: Confirm applicant identity.
Payslip income does not match the bank-credit average.
Next action: Request clarification.
An existing facility is missing from the application.
Next action: Recalculate affordability.
A signature is partially unreadable.
Next action: Request a clearer copy.
The latest bank statement is missing.
Next action: Request the document.
Illustrative example
Digital-first processing
Digital submission can improve data quality, reduce manual re-entry and create a clearer audit trail. Paper or handwritten forms can stay an optional exception workflow, with unclear fields routed for human verification rather than treated as confirmed data.
These are technical controls that can support the institution's privacy, security and data-handling requirements. Configured for the institution, they support PDPA-compliant handling of applicant personal data. They are not a certification or a guarantee of compliance on their own. Final compliance depends on the institution's policies, configuration, deployment and governance.
Indicative architecture
This is an example of how the layers can be assembled. The live tools follow the financial institution's environment. It is not a fixed production bill of materials.
Development partnership
Oxydata brings financial services and lending subject matter expertise, advanced AI modelling and go-to-market leadership. Nervesis focuses on software engineering and product development. Together, the teams are developing ZYGY Loan Approval Intelligence to support clearer, more consistent and traceable loan application reviews.
Oxydata
Nervesis
Talk to Oxydata about applying AI and workflow automation to your lending process, without standing up a separate scoring stack for every product.
Discuss a lending workflow