For community banks · BSA/AML alert triage
Your BSA team is good at finding fraud. They spend most of their time proving things aren't fraud.
We build the alert-triage system that filters the false positives, train your BSA officer to run it, and produce the SR 21-8 and SR 11-7 documentation the examiner will ask for. Your team works the alerts that matter.
The problem
A three-person BSA team processes hundreds of alerts a month. Sixty to eighty percent are false positives.
The investigators know within minutes that most of them are nothing. They still have to document each one. That documentation takes hours. Hours that don't go toward the alerts that matter — the patterns that actually threaten the bank.
Meanwhile the volume keeps rising. FinCEN reported 2.6 million SARs filed by depository institutions in FY 2024. SAR filings rose 18.5% between July 2023 and December 2024. The 2024 interagency/ICBA Payments Fraud Survey found 62% of institutions reporting a rise in sophisticated fraud tactics.
Your BSA officer knows all of this. She files the SARs, attends the exams, and manages the team. She also knows that the current transaction monitoring rules were tuned years ago, that the thresholds haven't kept up, and that the false-positive rate is quietly consuming her department's capacity.
The examiner doesn't see the false-positive burden. The examiner sees the SARs that were filed, the documentation that supports them, and the policies on the shelf. What the examiner cannot see is the fraud your team didn't have time to investigate because they were closing out alerts on a retiree's recurring church donation.
What changes
Your investigators focus on the alerts that deserve attention. The BSA officer owns the system.
After the engagement, your BSA team focuses on real alerts. False positives drop by 40 to 70 percent — consistent with what Verafin, Abrigo, and Unit21 have demonstrated at peer institutions. Pinnacle Bank achieved 30% fewer false positives and a 25% increase in prevented losses within 30 days. Abrigo's Texan Bank case: 70% false-positive reduction. MidCountry Bank went from thousands of alerts per month to hundreds.
Your investigators spend their recovered hours on the alerts that deserve attention. The ones with layered structuring patterns, unusual geographic flows, or velocity changes that signal real risk. Your team does more investigative work than they have done in years — because they are not buried in documentation that leads nowhere.
The BSA officer runs the system. She adjusts the thresholds. She reviews the model's recommendations against her own judgment and tunes the rules when the risk profile shifts. She does not depend on a vendor or a consultant to make changes. The system fits inside her authority, not around it.
SAR narrative drafting if you need it. If your bank files 200 or more SARs per year, the engagement includes SAR narrative drafting automation. The system produces a first draft from the alert data. Your investigator reviews, edits, and files. The narrative quality improves because the investigator spends her time on judgment, not formatting.
Governance documentation ships as standard. Model inventory, validation framework, and examiner-ready narrative. The work aligns with SR 21-8 principles for BSA/AML model risk and fits within the SR 11-7 framework your CRO already maintains. Your team knows where the documentation is because they helped build it.
How the build works
Two to four months. Four phases. Your BSA officer participates from the start.
The timeline depends on the complexity of your current transaction monitoring rules, the volume of alerts, and whether SAR narrative automation is included.
01
Audit
We review your current transaction monitoring rules, thresholds, and alert dispositioning patterns. We identify which rules generate the most false positives and why. Your BSA officer and at least one investigator participate — they know the system better than anyone.
02
Design
We evaluate vendor options and design a tuning program with backtested ML recommendations against your bank's own transaction data. The tuning is specific to your customer base, your product mix, and your risk profile. A rule set tuned for heavy CRE exposure looks different from one tuned for a large retail deposit base.
03
Build and teach
We build the system, integrate it with your existing transaction monitoring platform, and train your BSA team to run it. Your investigators learn how the model makes recommendations. Your BSA officer learns how to adjust thresholds and evaluate model performance over time. She owns the system when we leave.
04
Governance
We produce the SR 21-8 and SR 11-7 documentation — model inventory entry, validation plan, effective-challenge framework, and the examiner-ready narrative. Built alongside the system so the documentation reflects what was actually deployed.
What your team learns. Your BSA officer learns to evaluate model performance, adjust thresholds, and maintain governance documentation as the risk environment changes. Your investigators learn what the model does and why, so they can exercise informed judgment on every alert. Your CRO receives documentation that fits the model risk management practice she already oversees.
Vendors we evaluate. Nasdaq Verafin, Abrigo, Unit21, Feedzai, DataVisor — and whatever your bank is already running. We are vendor-neutral. We recommend what fits your bank's scale, risk profile, and existing infrastructure.
Start with the diagnostic
Book a 90-minute working session.
The diagnostic tells you where your BSA team's time goes, which rules generate the most false positives, and what a tuned system would look like at your bank. Your BSA officer participates. She leaves with a clear picture of what changes, what stays, and what the governance looks like.
You describe the problem. We describe how we would approach it. You leave with a written assessment within 48 hours.