Remittance and cross-border payment risk plays out across sending and receiving institutions in different jurisdictions. Chefoba was built and evaluated on exactly this kind of problem.
Join the waitlistWhere this came from
Chefoba's federation core was developed and evaluated as a 30-day remittance transaction prediction system across two UK fintech platforms, partitioned into four platform and geography clients covering 21,458 users. This wasn't a synthetic benchmark. It reflects the shape of a real cross-border payments problem: multiple platforms, multiple geographies, and payment corridors with materially different risk profiles.
One of the clearest findings from the evaluation: risk and model performance are not uniform across payment corridors. A model can look accurate in aggregate while performing meaningfully worse on specific corridors, for example a GBP-denominated corridor to a specific receiving market. Subgroup auditing found a 0.797 true positive rate gap between corridors in the evaluation, invisible in top-line accuracy alone.
If your institution operates across multiple payment corridors or receiving markets, an aggregate risk model can quietly underperform on the corridors that matter most for your exposure. Chefoba's subgroup auditing is designed to catch that before it reaches production, and its trust-weighted aggregation lets corridor-specific institutional knowledge contribute proportionally to the shared model.
Note on this evaluation
The findings referenced here come from Chefoba's own research evaluation, not a live client deployment. We describe the corridor and platform findings qualitatively rather than naming the institutions involved.
We're onboarding a limited group of fintech and banking partners before general availability.
Join the waitlist