Your model was validated by the firm that sold it.
Your screening tool, monitoring system or agent was validated by the company that sold it to you. We test it against your own data — above-the-line and below-the-line, thresholds, scenarios, autonomy boundaries — and write the model risk file.
We do not validate agents we built, and we hold no reseller agreement with any vendor we test.
We do not validate agents we built.
We hold no reseller agreement with any vendor we test. Every engagement is independent — from the scope letter to the model risk dossier.
If the model passes, you have a defensible file. If it doesn't, you know before the examiner does.
- Conflict-free — no financial relationship with any vendor we test
- Examiner-ready output — model risk file named, dated, and structured for OCC, FDIC, and NCUA
- Your data, not synthetic — testing runs on your live or historic transaction set
- Full scope — above-the-line and below-the-line, not just headline accuracy
Six dimensions, one dossier.
Threshold & alert accuracy
Are the thresholds set correctly for your customer population? Are alerts firing where they should and suppressed where they shouldn't? We test across transaction type, channel, and customer segment.
Scenario & typology coverage
What does the model miss? We run known typologies — structuring, layering, trade-based ML, correspondent-bank exposure — against your data and document the gaps.
Human-in-the-loop verification
For agentic systems, we map every decision point: what the agent decides, what it escalates, and what requires human sign-off. Boundaries are documented and tested, not assumed.
Examiner-ready model risk dossier
A complete SR 11-7 compliant model risk management file: model inventory entry, validation methodology, test evidence, findings, and compensating controls where gaps exist.
Four to six weeks, fixed price.
What we validate.
Sanctions screening, PEP and adverse media, name-matching systems and fuzzy-logic engines.
Transaction monitoring platforms, rule-based and ML-based alert generation, segmentation models.
Compliance agents, LLM-powered case analysis tools, AI-assisted SAR and CTR drafting systems.
Know what your model does before the examiner does.
A 30-minute scoping call. We will tell you what the engagement covers, what it costs, and how long it takes — before you commit to anything.