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.
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
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
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.
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