← Showcase solutions
Banking · Concept showcase

Loan Origination Assistant

A prototype showcase demonstrating how an AI underwriting copilot could support risk assessment, affordability analysis, and audit-ready decisioning — taken through Foundry: planned in Genesis, built in Forge, certified by Guardian.

Explainable
Decision rationale
Auditable
Evidence per decision
Minutes
Not days, by design
app.loanassist.ai / applications / LN-20471
SECURE
APPLICATION · LN-20471

Maria Hoffer

Mortgage · €340,000 requested · Belgrade
AI RECOMMENDATION: APPROVE
Application Score ▲ 12
742/850
Excellent
Risk Score Low
86
Low risk
Approval Probability
72%
Confidence 0.94
Recommended Amount −€0
€340,000
LTV 78% · 25y · 3.9%

Decision Factors

SHAP contribution
Credit history +0.42
Income stability +0.31
Debt-to-income -0.18
Collateral value +0.24
Employment length +0.15
Recent inquiries -0.09
Savings buffer +0.12

Affordability

monthly · EUR
€6,500Income
€4,450Obligations
€2,050Surplus

AI Underwriting Summary

Applicant presents a strong, stable profile. Credit history F1 and income stability F2 are the dominant positive signals. Debt-to-income F3 sits within policy at 34%. Affordability analysis confirms a €2,050 monthly surplus after the proposed loan. Recommend approval at full requested amount with standard terms.

Applicant Profile

Age / Tenure38 · 11 yrs employed
Annual income€78,000
Credit bureauA− · 0 defaults
Debt-to-income34%
Collateral LTV78%
Policy checks14 / 14 pass

Comparable Applications

last 30 days
Application Score Risk Trend Decision
LN-20471 742 86 Approve
LN-20455 688 71 Approve
LN-20448 604 58 Review
LN-20431 551 39 Decline
LN-20420 715 79 Approve

Concept showcase · all data shown is illustrative, not from a real applicant.

Problem

Underwriters spent days manually assessing each application across fragmented systems. Decisions were inconsistent, slow, and hard to defend to auditors and regulators.

Solution

A risk engine and underwriting copilot that ingests application data, scores creditworthiness, simulates affordability, and produces an explainable recommendation with cited factors.

Architecture

Event-driven services on the engineering platform, an ML scoring layer, and a policy-as-code compliance gate — all generated and governed through the Software Factory pipeline.

Intended Outcomes

Decisions in minutes rather than days, more consistent policy application, and audit-ready evidence emitted automatically for every decision — the outcomes this approach is designed to deliver.

Build a copilot like this on your data