Mandate Labs
use case aml and sanctions

Review the account, not the alert.

Upload a statement, a card export or your alert queue in whatever layout your systems produce. Ward reads it, flags what matches a typology, screens every counterparty and hands your analyst the evidence. Ward files nothing.

[02] For the model risk team

Built to be examined.

A compliance model that cannot be explained to an examiner is a liability. Ward's governance pack is written against US model risk guidance.

  • Model card and inventory. Purpose, components and versions, intended use, data, performance, limits and change control.
  • Validation report. Performance against analyst feedback and outcomes, score stability and latency, mapped to SR 11-7 and OCC 2011-12.
  • AML is advisory by default. The AML line informs the decision; it gates only if you turn that on.
  • Lists checked, every time. A list that was not loaded is "not checked", never "clear".
  • Calibrated on reference data. The statement AML line has a held-out calibration error of 0.039. It is refitted on your outcomes once you have them.

See Ward on your own cases.

Bring a statement, a declines export or the case your team is stuck on.