AML AI assistant

Introduction

A screening run against sanctions, PEP and adverse media lists returns hits, and most of them
are not your customer — a namesake, a different date of birth, a person in another country.
Reading each one to establish that is where AML review time actually goes.

The AI assistant reads them first. It qualifies each hit as a true positive, a false positive,
or one that needs a human, and writes the comment explaining why. Over 90% of hits come back
pre-qualified, so your analysts open the ones where the answer is not obvious and confirm the
rest in bulk.

The decision stays yours: nothing is closed automatically, and the AML check is only approved
once a reviewer validates it.

How it works

Every hit gets one of three qualifications, and each comes with a written justification you can
keep, edit or replace:

The qualification is a recommendation, not an action — the hit keeps its status until someone
applies one.

Reviewing hits

Filter the hit list by AI qualification, and by list type — sanctions, adverse media, and the
others your screening configuration covers. In practice you work the manual-review hits first,
then clear the rest.

Once a list is filtered, Change status to applies one qualification to everything selected,
and Discard closes the false positives in one action. The AI's comments are prefilled, and
Apply to all empty review comments fills any hit you have not commented yourself — so the
audit trail is complete without typing the same sentence forty times.