Case Study: Airvantage achieves 10,000 more advances approved per day and 15% fewer rejections with H2O.ai Driverless AI

More Accurate, Real-time Risk Score with Fast Time-to-Market

Airvantage, a South Africa–based telecom provider of the Prepaid Airtime Advance System (PAS), faced a business challenge where static, rule‑based risk scoring was overly defensive and producing false negatives—declining desirable customers and leaving revenue on the table. After evaluating over 30 toolkits and running multiple PoCs, Airvantage selected H2O.ai’s Driverless AI (with MOJO deployment) to build a data‑driven behavioural risk model.

H2O.ai’s Driverless AI sped feature engineering and model development from weeks to hours and the team moved from evaluation to production in one month, using MOJO for fast on‑premises deployment and live A/B testing. The new H2O.ai model augments the rules engine and delivers millisecond scoring while enabling Airvantage to approve about 10,000 more advances per day, cut advance rejections by 15%, and maintain repayment rates that exceed targets.

Airvantage

Marom Mishan

Actuarial Data Scientist