Payment Screening Agent: Accurately Clearing False Positives & Flagging More Risk
Alviere's challenge: rising screening alerts and false positives
- Alviere was experiencing a rising volume of payment screening alerts and had begun evaluating agentic solutions to help manage the load
- Accuracy was non-negotiable: any agentic solution needed to ensure no suspicious activity was missed while keeping false positives to a minimum
- Alviere decided to test Hawk's Payment Screening Agent against other available options to determine which delivered the greatest accuracy
- Hawk's Payment Screening Agent was run against a set of recent screening alerts, so Alviere could accurately compare solutions
Results in summary from the Payment Screening Agent
- Hawk's Screening Agent correctly escalated confirmed watchlist matches
- The Hawk Agent also correctly cleared all confirmed non-matches
- Most importantly, the agentic system provided a more comprehensive disposition narrative for Alviere's compliance team
How the Payment Screening Agent was tested
- Tested Recent Screening Alerts: Alviere tested Hawk's Screening Agent on recent counterparty screening alerts for potential sanctions matches. The Agent combined hit evidence with customer KYC data to confidently assess each alert.
- Identified the Party Behind Each Alert: The Agent first determined who the alert related to: the account holder or the transaction counterparty. Account-holder hits were checked against KYC details, while counterparty hits were checked against the transaction's counterparty information only to
avoid hallucination risks. - Applied Strict Match Standards: The Agent applied the same match standards an analyst would. A match required the party's name or a verified identifier, such as date of birth, address or registration number, to correspond to the listed entity.
- Escalated Unresolved Matches: The Agent's assessment was based only on screening list evidence and Alviere's internal data. Where the evidence supported a clear affiliation, the Agent resolved the alert. Where the evidence was insufficient to confirm or rule out a match, the Agent flagged it for human review.
Results: from deployment to results in two weeks
Alviere ran a set of recent counterparty screening alerts through Hawk's Payment Screening Agent to compare its decisions with an alternative agentic AI provider. The Hawk Agent confidently cleared 96% of the alerts and got every case right.
For the remaining 4%, it recognized that the evidence wasn't sufficient to make a confident decision and flagged the alerts for human review. This was a key difference from the alternative provider, which missed several of the alerts the Hawk Agent flagged for review.
Alviere: scaling a business with AI for financial crime and compliance
Alviere works with enterprises across a broad range of markets and customer segments, managing regulatory risk across different customer types, business lines, and transaction behaviors.
As more enterprise customers onboarded and transaction volumes grew, so did the risk of illicit payments slipping through screening. To stay ahead, the team expanded its screening rules to reduce the chance of genuine risk being missed. But broader screening came at a cost: far more alerts entered the
review queue, most of which turned out to be false positives. Each alert still required a manual review before the team could close it.
To reduce manual effort and lower the cost per screening case, Alviere needed a more efficient way to triage alerts. Hawk's Payment Screening Agent helps Alviere do this by analyzing hit evidence and customer KYC data, determining who an alert relates to, applying strict match standards, and escalating unresolved matches for human review.
Read more about the Hawk Payment Screening Agent