1 of 5

AI-POWERED LOAN INTELLIGENCE

CrediLume

Know Before

You Borrow

Team: KAILAH · Team_Lead: Dev Agrawal

THE PROBLEM

Why do 42% of loans get rejected?

No Eligibility Clarity

Borrowers apply blindly. Banks reject 4 in 10 applications — wasting time, damaging credit scores, and causing financial anxiety.

Zero Risk Transparency

Existing tools only calculate EMI. None explain WHY a loan is risky or which financial factors are hurting approval chances.

Farmers Are Excluded

86M Indian farmer households with seasonal income are rejected outright by fixed-salary-only credit models.

2 of 5

SOLUTION + HOW IT WORKS

One platform. Complete loan intelligence.

ML Approval Prediction

Random Forest model gives approval probability before you visit a bank. No more guessing.

Explainable AI

Shows exactly which factors (income, DTI, credit score) are helping or hurting your chances.

Scenario Simulation

Adjust income, loan, or tenure and see approval odds change in real-time.

Farmer-Inclusive Model

Yield × Price × Land = income. Seasonal EMI aligned to harvest cycles.

HOW IT WORKS →

1

Enter Details

Income · Loan · Debt · Credit

2

ML Analysis

10+ features · Random Forest

3

Risk Score

Probability · Low/Med/High

4

Insights

Top factors · Safe EMI range

5

Simulate

Tweak inputs · Improve odds

3 of 5

MARKET OPPORTUNITY

Massive, underserved, and ready.

600M+

Underserved borrowers in India

₹22L Cr

India retail credit market 2025

42%

Average loan rejection rate

86M

Farmer households needing credit

TARGET SEGMENTS

Salaried / First-Time Borrowers

Young professionals taking home or car loans

Want confidence before entering a bank

Fear rejection damaging their CIBIL score

Self-Employed / SME Owners

Variable income misread by static models

Often over-rejected despite good cash flow

Need a tool that reflects true affordability

Farmers (Kisan Module)

Seasonal yield × market price = income

EMI aligned to harvest cycles, not months

First platform to credit seasonal earnings fairly

4 of 5

TECH · MVP · BUSINESS MODEL

What's built, how it works, and how it makes money.

TECH STACK

ML Model

Random Forest + CatBoost+SHAP · 85-89% accuracy

Backend

Python · Flask / FastAPI · scikit-learn

Frontend

React.js +HTML + JS · real-time state updates

XAI

Feature importance → user-readable reasons

Kisan Engine

Yield × Price × Land → seasonal EMI

WHAT'S BUILT

Approval probability prediction

Risk classification (Low/Med/High)

DTI + affordability check

Explainable AI factor breakdown

Scenario simulator · Farmer module

BUSINESS MODEL

B2C Freemium

Free checks. Premium ₹99/mo unlocks scenario simulator + full report.

B2B API

Banks & NBFCs pay per-query or monthly SaaS to embed pre-approval.

Affiliate Leads

Earn commission per loan referral. Pre-vetted users convert 3–5× better.

Data Insights

Anonymised borrower behavior sold to BFSI research & rating agencies.

5 of 5

WHY WE WIN + ROADMAP

We don't just calculate — we illuminate.

First-Mover

No Indian platform combines ML prediction + Explainable AI + farmer inclusion in one tool.

Trust Engine

Transparent decisions build loyalty — users return when they understand the 'why'.

Dual Moat

ML improves with each user. Farmer module needs deep domain knowledge to replicate.

API-Ready

Cloud-first architecture means one integration scales to millions of queries instantly.

LAUNCH ROADMAP

Phase 1 NOW

· Public web MVP · Free B2C tier

· Social distribution to students / young pros

· Feedback loop + model accuracy tuning

Phase 2 3–6 Mo

· API pilot with 2-3 NBFCs

· Premium tier + affiliate bank partnerships

· Expand rural kisan user onboarding

Phase 3 6–18 Mo

· Full banking data integration

· KYC + real-time credit bureau pull

· Series A fundraise + national scale

"Every loan decision deserves intelligence, not guesswork." — CrediLume