Work the case that matters first.
Cases are ranked by what each costs if ignored, with the evidence behind them and the places where sources disagree. Holds, reviews and step-up checks each carry a price, so the false-positive trade-off becomes a decision someone can review.
In plain termsCases ranked by money at stake, with the evidence and where it disagrees.
At a glance
Ranked by money at stake
The queue is ordered by value, not by how unusual a case looks.
Conflict kept in view
Where sources disagree the case is marked contentious, not averaged away.
Priced actions
A hold, a review or a step-up check, each with its cost.
A defensible record
Evidence, decision, actor and time on every case.
The questions on a risk leader’s desk
- Which of thousands of alerts matters?
- What are our false positives costing, and who is pricing them?
- Where do our sources disagree, and does anyone see it?
- Could we defend this decision to a regulator?
One case, start to finish
A case is ranked with its evidence, an analyst approves a priced action, and the evidence, decision, actor and time go on the audit trail.
One case, start to finish
- 1. Account activity to AnantState: Behavior breaks from its history
- 2. AnantState to An analyst: A ranked case, sources in conflict
- 3. An analyst to AnantState: Approves a priced hold
- 4. AnantState to Audit trail: Evidence, decision, actor, timea reply or asynchronous
Why the disagreement matters
A weighted score averages two weak sources and two flatly contradicting ones to the same number. Evidence fusion keeps the contradiction, because a contested case is the one an experienced analyst should look at.
Same average, different case · illustrative
| Category | Source A | Source B | Result |
|---|---|---|---|
| Weak agreement | 0.5 | 0.5 | Low conflict |
| Contradiction | 1 | 0 | High conflict |
Concerns a risk leader is likely to raise
| The concern | The answer |
|---|---|
| Model risk will want to validate it. | The model record is built for that: documented purpose, performance against a defined baseline, a recorded promotion decision and drift monitoring. |
| We can’t let a system act on contested evidence. | It doesn’t. The conflict gate holds the action for a person regardless of the autonomy level. |
| Does it claim regulatory compliance? | No. We state the controls that exist and your compliance function maps them to the regulation. |
Bring a queue
Bring:
- A queue
- The rule nobody has changed in two years
You leave with the queue re-ordered by value at stake, and that rule priced against the loss it prevents.
Price the false positive
Bring a rule whose false-positive cost nobody has measured. We will price it against the cost of the loss it prevents.