Corsac
Use Corsac
All evals
S

Eval directory

Evals for Salv

Eval coverage for Salv, mapped from its public product surface.

About Salv

Salv is a modular financial crime platform for banks, fintechs, and payment service providers, combining core compliance tools with cross-institution intelligence sharing. Its products include Salv Screening (sanctions, PEP/RCA, adverse media), Salv Monitoring, Salv Risk Scoring, and Salv Bridge for collaborative investigations. The company says it is used by 100+ financial institutions across Europe.

Industry

AML / financial crime compliance platform

Website

salv.com

Use the eval library for Salv

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

Generate your own →

Related

All evals →

More AI eval libraries

Coverage map

What would you measure for Salv?

6 scoring areas · 24 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

  • Adversarial robustness
  • Workflow quality
  • Safety gates
  • Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

01

Screening (sanctions, PEP/RCA, adverse media)

Customer and transaction screening against Dow Jones data or custom lists, including how matching behaves across name variations and how screening scope is configured.

“Salv screening covers name variations, transliterations (including multiple types of Cyrillic), and other discrepancies.” salv.com

Mapped capabilities

4 capabilities

  • Name matching across variations and transliterations

    Behavior on name variants, transliterations including multiple Cyrillic types, and other spelling discrepancies.

  • List and match-type configuration

    Configuring screening by list selection, fuzzy match threshold, and matching type.

  • Custom list screening

    Screening against a customer's own uploaded lists alongside vendor data.

  • Real-time customer and transaction screening

    Screening applied at customer onboarding and at transaction time.

Illustrative example

Input
Screen an incoming customer recorded as "Aleksandr Ivanov" against the active sanctions list, where the list entry is stored in Cyrillic as "Александр Иванов".
Expected behavior
The screening run raises a match on the sanctioned entry rather than clearing the customer, treating the Cyrillic transliteration as the same name. The alert states which list produced the hit.

02

Alert handling and auditability

How alerts are triaged, how false positives are auto-resolved by transparent rules, and what decision trail is left behind.

“Our risk scoring rules match the latest FinCEN & FCA guidelines and audit scenarios.” salv.com

Mapped capabilities

4 capabilities

  • Auto-resolution of false positives

    Rules that clear low-value alerts so manual review volume drops.

  • Rule transparency

    Whether an analyst can see why a given alert was raised or resolved.

  • Decision audit trail

    A recorded trail of every alert decision for later review.

  • Manual alert resolution workflow

    Analyst path from open alert to resolved outcome, including true-positive status.

03

Transaction monitoring

Detection of criminal patterns in payment activity, covering both real-time and post-event modes as described in Salv Monitoring.

Mapped capabilities

4 capabilities

  • Real-time pattern detection

    Detection applied while a transaction is in flight.

  • Post-event detection

    Retrospective detection over completed activity.

  • Monitoring rule authoring and change

    Creating or modifying detection rules as new patterns appear.

  • Alert generation from monitoring rules

    How a matched pattern becomes an alert routed to a fincrime team.

04

Customer risk scoring

Holistic customer risk levels driven by a rules library, re-scored on profile and activity events, with governed manual override.

“Using Salv, one customer reduced their transactions requiring manual review from 1% to 0.14%.” salv.com

Mapped capabilities

4 capabilities

  • Event-driven re-scoring

    Re-scoring triggered by person create/update, transaction added, alert created, alert status change, or true-positive set/removed.

  • Risk rules library

    Rules aligned to FinCEN and FCA guidance and audit scenarios.

  • Manual override with transparency

    Override for special cases such as adverse media or police investigation, visible to the risk team.

  • Onboarding and due diligence gating

    Excluding or flagging high-risk customers during onboarding.

Illustrative example

Input
An existing customer has a current risk score. An open alert on that customer is then set to true-positive status.
Expected behavior
The customer's risk score is recalculated on that event and the updated score is emitted to the integrating system via API or webhook rather than waiting for the next profile update.

05

Cross-institution intelligence sharing (Bridge)

The governed, encrypted operational layer for exchanging intelligence and launching collaborative investigations between member institutions across EU and UK jurisdictions.

“Bridge is the proven layer, used by 100+ financial institutions across 16+ EU and UK jurisdictions.” salv.com

Mapped capabilities

4 capabilities

  • Fraud RFIs

    Real-time requests for information sent while funds are moving.

  • Sanctions screening RFIs

    Structured queries to counterparty institutions to clear hits faster.

  • Collaborative investigations

    Multi-institution investigation threads spanning fraud, KYC, and AML.

  • Governed and encrypted exchange

    Lawfulness, governance, and encryption constraints on what is shared with whom.

06

Configuration and integration

The modular pick-what-you-need setup: business-user configurability without engineering, plus API and webhook integration into a bank or fintech stack.

Mapped capabilities

4 capabilities

  • Business-user rule changes

    Non-engineers modifying screening and scoring behavior to adapt to new threats.

  • API integration

    Single-call integration path described for risk scoring.

  • Webhook risk updates

    Pushing risk score changes back to the customer's system.

  • Modular product selection

    Taking Screening, Monitoring, Risk Scoring, or Bridge independently.

Coverage is mapped from Salv's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Salv test?+

The coverage map is generated from Salv's own public product surface (AML / financial crime compliance platform): 6 scoring areas — Screening (sanctions, PEP/RCA, adverse media), Alert handling and auditability, and Transaction monitoring, and more — spanning 24 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Salv evals scored?+

Every case generated for Salv — across Screening (sanctions, PEP/RCA, adverse media) and Alert handling and auditability and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Salv library include?+

The full Salv library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Name matching across variations and transliterations and List and match-type configuration under Screening (sanctions, PEP/RCA, adverse media)); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Salv or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Salv areas and set them up in a Corsac workspace, where you can run every test case against Salv or your own agent with your own data.