Repeated
the same clearance, every week
The cause of last month's clearance was never identified. The same match is returned and is processed from zero.
Agentic AI · Screening
Agentic AML-Risk Ops
Native to the platform, trained on your own case history. The agent reads every screening and monitoring alert, calls it a false positive, a true match, or human review, and acts on the rule you draw on the canvas.
Minutes
From idea to live rule
No code
Drag, drop, connect
Dry run
Tested before it acts
Every run
Written to the log
Why this matters
Alerts don't get better with more reviewers; they need a decision that remembers what was decided.
Most alerts have already been dealt with; when the name, list, and result are the same, someone else dismisses them again the following week. Your team already knows the answer because the agent learns from how you closed previous cases and applies it, so analysts end up working on the cases that actually need their input.
Uneven
one threshold per analyst
Every reviewer draws their own boundary between the suspicious and the safe, and that boundary shifts as the situation changes.
Unwritten
no trail behind the decision
When an examiner is asked why an alert was dismissed, the rationale is stored in memory, not in the file.
How it works
Scroll to advance
There are four nodes. The queue empties on its own.
-
Step 01
The alert lands on the customer record
Next to all the other information already known about that customer, each screening or monitoring hit includes the match rate, the list source, and the entity type.
-
Step 02
The agent refers to it in the manner that your team would
It is trained in the way your team has previously closed cases and then gives a result which is either a false positive, a true positive, or an escalation, together with a confidence score.
-
Step 03
Confidence sets the route
If the confidence is above 90, close the case. When the number is between 70 and 90, the operator takes it up with the agent's note. If it is under 70, escalate it.
-
Step 04
The action takes place and is recorded.
Close, set match status, assign, tag, note or trigger a webhook. Each result is immediately recorded in the log.
sanctionscanner / alerts
- Open alerts1,284
- Auto-eligible912
- Median age2 min
sanctionscanner / agent / decision
- DecisionFalse positive
- Confidence94
- BasisDOB + nationality
sanctionscanner / rule / QNB_Rule
- If confidence> 90
- ThenClose case
- ElseAssign operator
sanctionscanner / agent / log
- ResultApplied
- RuleQNB_Rule
- Retained5 years
Read
Decide
Branch
Act
The rule builder
One rule, drawn once
The full rule on one screen: a trigger, a confidence condition, three outcomes. Not a line of code.
What the builder gives you
Rules that read like plain sentences
Visual rule canvas
Place a trigger, a condition, and an action, then connect them. No script and no release cycle.
43 condition fields
Go by the confidence score, the match status, the risk level, the entity type, the amount, the channel, the number of articles, and the like.
11 operators
Equals, greater than, between, in, is true, and more. You can add conditions just as you would say them.
13 actions
Close, set status, set risk, assign, tag, note, notify, convert to case, send a webhook, switch monitoring on or off.
Branch and switch
Several outcomes on one canvas. Clear matches close, uncertain ones route, the rare hard ones escalate.
Dry run before live
Apply the rule to the live alerts within the simulation, and you'll see all the decisions it would have reached before any case moved.
What it leaves behind
Nothing runs off the record
An agent without a trail is an audit finding waiting to happen. Each run records what it read, what it decided, and what it changed. The agent works as its own user in your account, so every step is attributable.
Applied
Rule matched, action completed
Logged
Dry run
Simulated, nothing changed, ready to measure
Logged
Skipped
Rule matched, a guard held the action
Logged
Failed
Action did not complete, reason recorded
Logged
FAQ
Before you switch it on
What does the agent do with a screening alert?
It reads each screening and monitoring alert on the customer record, with the match rate, list source and entity type, and labels it a false positive, a true match or an escalation. It then acts on the rule you drew on the canvas, and every step is logged.
How does the agent learn our team's decisions?
From how your team closed previous cases. When the name, the list and the result match a decision your analysts have already made, the agent applies that decision instead of sending the alert back into the queue. Your team's judgement becomes consistent, not replaced.
What happens when the agent's confidence is low?
Every decision has a confidence score, and the score sets the route. High confidence closes, the middle band goes to an operator with the agent's note, and low confidence escalates. You set the bands, for example above 90 to close and below 70 to escalate.
Can the agent close alerts without an analyst?
Only where your rules let it. You can keep any rule in dry run for as long as you want, measure what it would have closed, and switch it on only when the results satisfy your team and your auditor.
Which actions can the agent take?
Close, set match status, set risk, assign, tag, add a note, notify, convert to a case, trigger a webhook, or switch monitoring on or off. Each action is recorded against the rule and node that triggered it.
Can our compliance team build a rule without code?
Yes. A rule is a trigger, a condition and the actions to take, connected on the canvas without code. Conditions draw on 43 fields, such as confidence score, match status, risk level and entity type, and read like plain sentences.
How do we know a rule works before it goes live?
Run it in dry run first. Dry run applies the rule to live alerts in simulation, so you see every decision it would have made before any case moves. You can compare those decisions with your analysts' own and adjust the rule until they agree.
Can an examiner see what the agent did?
Yes. Every run logs which rule and node ran, what the agent saw, what it changed, and whether the action was applied, simulated, skipped or failed. A failed action carries its reason, and a skipped one shows which guard held it.
Are the agent's actions separate from our analysts'?
Yes. The agent works as its own user in your account, so every step is attributable and separate from your analysts' work. An examiner can review the agent's decisions as a set.
How do we stop a rule immediately?
Set the rule to draft and it stops immediately. The log of everything it did stays in place for your auditor, and the rule can be edited, tested again in dry run and republished.
Bring a week of your own alerts
We will build one rule on the canvas, run it in dry run against that week, and show you what it would have closed.
Minutes
To first rule
No code
To change one
Dry run
Before it touches a case