Customer Risk Rating demo
Agents re-score a member the night a fact changes, explain the rating in one cited sentence, hold type-only and open-activity changes, and route every upgrade and downgrade to a maker and a checker 1:33
What the demo shows
BSA teams score each customer's AML risk from facts, explain the score and route changes
A Fact Watcher reads five systems each night (core banking, transaction monitoring, account opening, OFAC screening and case management) and links every changed fact to a member. The Rating agent scores the member under the approved methodology and writes the change as one sentence, each 'because' clause cited to its source and its weight. A Type-Bias Checker holds any High that rests on member type alone, and a Hold Checker holds downgrades during open activity and changes built on stale facts. Like changes go into batches with a risk-weighted sample. A BSA analyst approves, overrides or rejects with a reason code, and a different person signs second; moves into High hand off to enhanced due diligence. Methodology versions are back-tested on the whole book, validated by Model Risk and approved by the BSA Officer before use.
Transcript
- Customer Risk Rating scores each member's money laundering risk, explained in one sentence.
- Examiners expect it: no customer type is automatically high risk.
- Last night, Norrell Federal Credit Union's agents proposed 152 rating changes.
- Kestrel Ridge Supply moves to High: monthly cash rose from twenty-two to seventy-one thousand.
- Each because opens its source and its weight.
- A checker agent confirms type alone didn't decide it.
- Tamsin approves as maker; a second analyst signs, and due diligence is drafted.
- Leona Strick's downgrade is held: a joint account has an open alert.
- Tamsin rejects it with reason code open activity.
- Like changes go in batches: one decision for forty-one members.
- Each methodology version is back-tested on the whole book first.
- The result: every rating explained, and only 410 of 3,400 monthly changes read singly.
- Built on Botminds: agents, data, screens and governance on one platform.
The narration as written for the film. Captions are on the player.