Banking & Credit Underwriting

What this covers

• —Loan origination
• —Debt-to-income checks
• —Credit score policy evaluation
• —Deterministic decisioning with a step budget

Outcomes for this persona

What a successful evaluation looks like in production — not brochure metrics.

• —Same policy applied on every channel in milliseconds, not overnight batch
• —Explainable decline/approve with a full decision trail for auditors
• —Policy changes promote without an application redeploy

Loan origination purpose story

Separate credit policy from the loan origination system: authors own score bands, tenure gates and exposure caps as a decision table; channels call one published service.

1. Model the loan application contract — credit score, income, amount, employment years — as decision-table inputs (BOM/JSON vocabulary, not a Java XOM JAR product).
2. Author score-band and tenure rules with hit policy FIRST and an else path so every application gets approve, refer or decline with a reason code and max APR.
3. Publish a version and deploy to an environment (dev first). Hot/canary and blue-green are available when you promote further.
4. Execute from Decision Server / HTDS-style HTTP apps with a deterministic step budget; the same version answers every channel.
5. Inspect matched[] in Decision Warehouse — which rules fired, in order — for underwriters and auditors.

Trial seed: Free trial signup seeds Sample · Loan Pre-Approval (key loan-pre-approval): five rules, FIRST hit policy, published and deployed to dev in one transaction. The /demo loan scenario walks the same inputs and reason codes illustratively — the trial runs the real engine.