Beforepay Group launches Carrington Labs, a business line to commercialise its AI-powered risk models and lending platform

AI-powered paycycle detection and prediction - algorithms that use natural-language processing (NLP) and machine learning used to identify salary transactions and forecast income timing and amounts going forward, for income verification and repayment scheduling;

Data ingestion and pre-processing - a module to take in different types of alternative data, categorise transactions, and prepare the data for training and inference;

Advanced risk assessment - the centrepiece of this solution, our native AI risk-scoring model tests more than 50,000 individual variables to select the 400-500 most predictive variables that could impact a borrower’s financial stability and creditworthiness, including behavioural, financial, and other factors, using gradient-boosting algorithms;

Personalised limit-setting - an engine that automatically conducts programmatic A/B testing, randomising users into different cohorts with different limit journeys, to create data points which can then allow estimation of default elasticity. This enables the calculation of the value-maximising limit for each customer, based on likelihood of default at different potential limit sizes and the implications for expected customer lifetime value; and

Repayment scheduling and debiting - logic to combine the limit-setting outputs and the paycycle forecasting to create an individual, optimised schedule for customer repayments.