crosshairs-simpleCredit Scoring Engine

Credie’s Credit Scoring Engine is the core of its AI-based lending protocol. It replaces traditional credit bureaus with real-time, trustless, and transparent on-chain behavior analysis, giving borrowers a dynamic score based on how they act across wallets, protocols, and time.


🧠 How the Engine Works

At its core, Credie’s engine is an AI system that processes a user’s blockchain footprint, aggregates off-chain and social signals, and continuously updates their score using a hybrid trust model.

    A [Wallet Connect] --> B [Behavioral Data Extraction]
    B --> C[AI Risk Assessment]
    C --> D{Scoring Buckets}
    D --> E[Borrower Score]
    D --> F[Loan Terms]
    E --> G[Reputation Update]

πŸ“Š Score Composition

Credie Scores are broken down into weighted categories. Each factor is recalculated as activity changes, allowing for real-time credit evolution.

Factor
Weight (%)
Description

Wallet Behavior

30%

Historical DeFi usage, staking, liquidation events

Transaction Trust

20%

On-chain counterparties, smart contract risks

Repayment History

25%

Timely loan repayments across Credie + partner protocols

Social Graph + Verifications

10%

Web2 proof, GitHub/ENS/Twitter identity trust

Asset Diversity + Health

15%

Portfolio value, volatility, on-chain diversification


πŸ“ˆ Visual Breakdown


πŸ›°οΈ Connected Data Sources

  • βœ… On-chain: Solana, Ethereum, Arbitrum, Base

  • βœ… Off-chain: Lens, Twitter, GitHub (via Oracles)

  • βœ… Device metadata (optional): OS, geolocation clusters, screen resolution (used only for trust anchors)

  • βœ… Partner identity protocols (e.g. Gitcoin Passport, BrightID)


πŸ” Score Buckets

Each wallet is assigned a dynamic tier, recalculated weekly:

Tier
Range
Risk Level
Typical APR
Max Loan Amount

Prime

80-100

Very Low

3% - 6%

5,000 USDC

Preferred

60-79

Moderate

7% - 12%

3,000 USDC

Starter

40-59

High

13% - 18%

1,500 USDC

Limited

< 40

Restricted

20%+

500 USDC


βš™οΈ Example Use Case: Real-Time Scoring

Imagine a user:

  • Stakes on Solana for 6 months

  • Pays off 3 loans early

  • Links a verified GitHub with OSS contributions

Within 48 hours, the scoring engine will:

  • Boost their Repayment and Social buckets

  • Increase their loan cap by 30%

  • Lower their APR by 4% in select vaults


πŸ§ͺ Why This Matters

  • πŸ“‰ No middlemen: Credit scores are not owned by banks or companies, they’re owned by wallets.

  • πŸš€ Dynamic: Scores evolve with behavior. Your past doesn’t define your future.

  • πŸ”Ž Transparent: Users can audit exactly why their score changed.

Last updated