About Verifex

Screening decisions should be explainable.

Verifex helps teams screen sanctions, PEP, and entity-risk data while preserving match reasoning, list versions, algorithm versions, and reviewable evidence for every decision.

Why we exist

Sanctions screening is not enough. Every decision needs evidence.

Screening tools often return scores or matches. Compliance teams still need to explain those decisions later — to auditors, banking partners, and regulators.

Audits and internal reviews require more than a risk level. They require proof: which lists were checked, which versions were active, what the match reasoning was, and why the decision was recommended.

Verifex is designed around decision evidence, not only matching. Every screening produces a structured, retrievable record that preserves the full context of the decision at the moment it was made.

What we build

Infrastructure for reviewable decisions.

Sanctions & PEP Screening API

Multi-source screening against sanctions, PEP, watchlist, and debarment lists. Exact, fuzzy, phonetic, and contextual matching with structured confidence scores.

Evidence Capsules

Structured JSON records that preserve query input, list versions, algorithm version, match rationale, adjudication reasoning, and a deterministic hash for every screening decision.

Benchmark Transparency

A published benchmark test set with disclosed methodology, true positives, false positives, and edge cases. No sales call required to evaluate accuracy.

Sentinel-Tested Matching Engine

A continuously monitored matching pipeline that screens critical sanctioned entities across name variants, transliterations, aliases, and abbreviations — with regression testing after every change.

Decision evidence

Evidence Capsules preserve every decision.

Most screening tools return a risk level and discard the context. Verifex preserves a structured record — an Evidence Capsule — that reconstructs exactly why a decision was made.

Query input

The exact name, entity type, country, and date of birth submitted at screening time.

List version tracking

Which source lists were active, their ingestion timestamps, and version identifiers at the moment of screening.

Algorithm version tracking

The matching pipeline version and penalty configuration used to produce the result.

Match reasoning

Structured explanation of match stage, confidence derivation, penalty chain, and source context.

JSON evidence export

Export Evidence Capsules as structured JSON for integration into audit systems and compliance workflows.

Deterministic hash

Each capsule includes a tamper-evident hash so the record integrity can be verified later.

Accuracy

Matching accuracy should be measurable.

We publish a living benchmark with disclosed methodology, true positives, false positives, and edge cases. Anyone can evaluate Verifex accuracy without a sales call or an NDA.

Published test set

A vendor-authored benchmark covering exact matches, spelling variations, transliterations, phonetic matching, word-order changes, entity names, PEP screening, and adversarial inputs.

Disclosed methodology

How F1, precision, and recall are calculated — and what the benchmark does not cover.

False positive breakdown

Every false positive is categorized by cause and confidence level, with remediation status.

Open comparison

Results are compared against published research (OpenSanctions Pairs, Federal Reserve FEDS 2025-092) where applicable.

Matching engine

Regression-tested with sentinel entities.

The matching engine is continuously monitored against a set of critical sanctioned entities. If a code change causes a regression — a missed alias, a dropped transliteration, a false clear — it is caught before reaching production.

Name variant coverage

Full names, aliases, abbreviations, transliterations, and patronymic derivatives are tested across multiple scripts.

Cross-source validation

Critical entities are verified against multiple independent sources to reduce single-source blind spots.

Penalty pipeline

A 10-penalty scoring pipeline with contextual scoring, entity-type mismatch detection, and common-name disambiguation.

Continuous monitoring

Sentinel tests run on every deployment. Regressions block release.

Principles

How we build.

Evidence over black boxes

Every decision should be reconstructible. If you cannot explain why a name was cleared, the tool is not complete.

Explainability over opaque scores

Confidence scores are useful. Structured reasoning — match stage, penalties, source context — is necessary.

Regression testing over assumptions

The matching engine is monitored with sentinel tests for critical sanctioned entities. Changes are tested before they reach production.

Source transparency over hidden lists

We disclose which sources are checked, when they were last updated, and what version was active at screening time.

Developer-first infrastructure

REST API, structured JSON, webhooks, and clear documentation. No portal-first workflow that forces manual steps.

Compliance-safe wording

We provide decision-support evidence, not legal judgment or compliance guarantees. Final review is yours.

Decision-support, not legal judgment. Verifex provides decision-support infrastructure. It does not replace legal judgment, regulatory advice, or a compliance team's final review. Every screening result should be reviewed in the context of your own risk-based compliance program.

Get started

Build screening workflows that leave evidence behind.