AML compliance software: sanctions screening selection checklist

Define the job the software needs to do

AML compliance software can cover many tasks. For SanctionsKit, the task is selected-source screening and review: submit a customer, find possible matches, let an analyst review them and keep the evidence. Identity verification, comprehensive PEP or media research, payment analysis and regulatory filings need other services or processes.

Start with invented data in the workspace. Decide who will submit checks, investigate candidates and approve work. Set their permissions in Team & access. A developer's API key is a tool for the integration; it is not an analyst who can make human review decisions.

Start with a clear question for each step: who is the customer, which lists must be checked, and who decides what to do with a match? Use those answers to choose the controls your team needs.

Publish a policy your team can follow

An owner or administrator uses Screening policies to define the purpose, jurisdictions and sources. Choose required and optional sources by subject type, then record exclusions, outside checks, review rules and monitoring cadence. Confirm source eligibility before relying on it. Saving a source in a policy does not create access.

Give each outside check clear instructions and describe the evidence it needs. Publish the version and use it for test screenings. The policy guide explains saved versions and read-only API discovery. A later version records a change; it does not rewrite the scope of an earlier check.

Follow one request into the analyst queue

Use the API quickstart to send an authenticated synthetic request from your server. Keep the result ID and coverage. Open the retained case in Review queue. Confirm that the analyst can see the submitted customer, source facts and reasons for the candidate.

Test a potential_match, a no_match and a failed request. A no-match may still require outside checks or business approval under the policy. Read the current case before showing the customer as approved. The screening status alone cannot tell you whether the analyst's work is done.

Finish a review and check its evidence

Have a permitted analyst decide each candidate's identity, explain the business action, add evidence and finish outside checks. Submit the review. When independent approval is required, a different eligible person must approve it. Read the saved state after that approval.

Retrieve the screening evidence and the right case report. Open a linked original document at its intended version. External document references point to your approved storage service; they do not copy the original file into SanctionsKit. An export is useful only if the evidence it depends on remains available to the right people.

Assign the ongoing checks to an owner

Create monitors for relationships that need repeat checks, using the intended subject and coverage. Check the capacity forecast and then the first successful result. active means a monitor can be scheduled. It does not prove that a check has already finished.

Name who will handle failed sources, outdated customer facts and new evidence. Test Pause, Resume and an identity change in Sandbox. Use the monitoring checklist to check the process over time, and Usage & limits to plan its shared allowance.

Record what you verified before Production

Keep an acceptance record for the actual tasks you tested: source selection, an authenticated sandbox request, an open candidate, required approval, evidence retrieval and a failed check. Leave an outside dependency pending until someone has checked it. A ticked checklist without evidence is not a completed test.

Live checks need a free or paid allowance, available sources, the right permissions and approved handling of real data. Your first workspace gets 250 free live checks with no card needed. A sandbox test shows how the app works with made-up records. It does not prove live source accuracy or that every AML duty has been met.

Frequently asked questions