OracleAI

Credible's OracleAI is an AI-driven wallet scoring system that anonymously evaluates crypto creditworthiness using on-chain data, optimizing LTV ratios and enabling no-collateral lending.

Credible offers overcollateralized payment financing using stablecoins, with settlement in local fiat.

Our long-term vision is to transition from overcollateralized to undercollateralized, and eventually to no-collateral loans. Achieving this requires a reliable way to assess the creditworthiness of wallets, including their credit scores and histories.

To enable this, we are developing OracleAI—a system that stores credit histories based on various data points and uses AI to score wallets. This approach is similar to how traditional credit bureaus, like Experian and FICO, assess credit in conventional finance.

OracleAI is an advanced AI-powered credit scoring system developed by Credible, designed to assess and process creditworthiness using on-chain data. By analyzing financial behaviors across decentralized lending platforms, OracleAI generates dynamic, real-time credit scores that can optimize Loan-to-Value (LTV) ratios and facilitate more sophisticated underwriting decisions. This technology not only enhances collateralized loan models but, as it evolves, will also enable under-collateralized and even no-collateral lending, offering a new era of credit access within the DeFi ecosystem.

How OracleAI Works

OracleAI integrates an innovative node-based architecture, where multiple nodes across decentralized networks process credit scoring data and store it securely on-chain. These nodes aggregate data from decentralized finance (DeFi) platforms, providing real-time updates and reliable, transparent credit scores. Credible and other projects can leverage this data for critical credit assessments, such as determining LTV ratios, setting underwriting criteria, and making risk-based lending decisions.

Data Collection & Aggregation

OracleAI collects and processes data from various DeFi platforms, including Solend, Kamino, MarginFi, Banx, and Sharky, offering a comprehensive overview of wallet behaviors. The key data inputs include:

• Lending/Borrowing History: Tracks the wallet’s lending and borrowing activity, loan sizes, and frequency across platforms.

• Default Events: Identifies any historical defaults or missed payments, contributing to the risk profile.

• Liquidation Data: Monitors past liquidations due to undercollateralization, providing insight into the wallet’s risk tolerance.

• On-Time Payments: Assesses the wallet’s consistency in meeting repayment deadlines for principal and interest.

• Risk Exposure: Evaluates the wallet’s risk level based on leverage and collateralization across multiple DeFi platforms.

These data points are processed through OracleAI nodes, ensuring that credit scores reflect real-time financial behaviors.

Scoring Model

OracleAI employs a sophisticated scoring model that combines both quantitative and qualitative factors, resulting in a comprehensive credit score. Key metrics include:

• Credit Utilization: Measures the ratio of borrowed funds to available collateral. High utilization indicates higher risk.

• Repayment History: Tracks on-time payments and overall repayment behavior, a key metric in determining trustworthiness.

• Liquidation Rate: Monitors the frequency of forced liquidations. A higher liquidation rate indicates higher risk.

• Risk Exposure: Evaluates the level of exposure to high-risk, leveraged positions across platforms. More leverage results in a lower score.

• Credit History: Assesses the wallet’s complete history of borrowing, lending, and repayment activity across all platforms.

These metrics generate a credit score, ranging from 0 to 1000, where higher scores indicate greater creditworthiness and reliability.

Machine Learning & Real-Time Updates

OracleAI utilizes machine learning models like Random Forests and Gradient Boosting to predict future credit risk based on real-time data. The system continuously updates credit scores as new transactions are processed, ensuring that scores remain accurate and relevant, reflecting the latest financial behavior of the wallet.

LTV Optimization & Future Lending Products

Initially, OracleAI enhances Loan-to-Value (LTV) ratios for collateral-backed loans by providing more precise and data-driven credit assessments. As the AI system matures, it will enable under-collateralized lending and, eventually, no-collateral loans, democratizing credit access in DeFi.

• LTV Optimization: By using OracleAI’s data, lenders can make more informed decisions about acceptable LTV ratios, minimizing risk.

• Future Lending Models: As OracleAI’s scoring engine evolves, it will facilitate more flexible lending models, including under-collateralized and unsecured loans, expanding the pool of accessible credit.

Leveraging On-Chain Credit Data for Broader DeFi Adoption

The OracleAI nodes and credit scoring system are designed to be used not only by Credible but also by other projects within the DeFi ecosystem. Through on-chain credit data and transparent scoring, DeFi platforms can integrate OracleAI to improve lending decisions, reduce risk, and expand access to credit for users with limited collateral.

• Cross-Platform Credit Scoring: The data collected and processed by OracleAI is available on-chain, creating a unified, decentralized credit score that can be accessed by multiple DeFi protocols.

• Underwriting Decisions: With more accurate, real-time credit scores, projects can optimize their underwriting models, reducing reliance on traditional credit scores like FICO.

• Decentralized & Secure: Data processed through OracleAI nodes is immutable, ensuring transparency and trustworthiness for users and projects alike.