Outputs / Case Studies
Choose a study, then explore its reports.
Real public retail data. Customer economics, demand patterns and a separate engine-ingestion rehearsal. Descriptive insights, not causal effects.
A synthetic payments benchmark. Explore the engine's test-run results and their limitations, not a production fraud-accuracy claim.
APP fraud reporting over a 494-send book. Explore the executive dashboard and regulatory pack. Detections, not measured interventions; no accuracy or causal claim.
When should you trust an AI alert? Three experiments with public chess data explore what unusual patterns can tell you, where models fail and what evidence should justify action.