Compared to traditional machine learning models, NeoSpace’s foundation models powered by NVIDIA delivered better results across customer populations, data modalities, and over time. NeoSpace achieved at least 30-50% higher accuracy, enabling its financial services customers to expand AI use cases without re-architecting their infrastructure.
In a landmark deployment, NeoSpace collaborated with another major Latin American financial institution to create a propensity modeling solution on large-scale tabular data to identify customers most likely to take out installment credit.
The model was evaluated using the Kolmogorov–Smirnov (KS) test, acceptance rate and conversion rate by decile. This deployment reached 6 million production customers and leveraged over 10TB of data covering over 50 million including CRM Feature Stores, eligibility variables, historical contract behavior, transactional data and much more.
Compared to the bank’s existing reference resources, NeoSpace improved KS performance by 10 to 20 points. The acceptance rate in the top decile increased by 20 to 30 percentage points, showing that a greater proportion of customers who signed up for the bank’s installment credit offering were concentrated in the higher propensity segment. These performance gains allowed the bank to contact fewer customers while achieving higher overall results, increasing effectiveness per interaction and enabling higher impact campaigns.
Overall, the improved propensity strategy resulted in a 10% increase in installment credit offerings, driven by better ranking of high-propensity customers and accurate engagement. These foundation models constitute the institution’s new core predictive layer, which provides more stable predictions, faster time to value, and improved business efficiency.
