Anti Money Laundering & Risk Intelligence

Detect risk before it becomes loss.

Machine learning and rule-based monitoring — built for SACCO compliance and audit teams.

Risk Intelligence

It knows what normal looks like — for every single member.

AuditPulse builds a private behavioural profile for every member, account, and staff user — learning typical transaction amounts, usual times of day, and familiar counterparties over time. Every new transaction is compared against that person's own history, not a generic threshold, so a KES 80,000 transfer is only flagged when it's genuinely out of character. Rising risk is tracked over time, not just moment to moment, surfacing members on a watchlist before a single transaction ever crosses a hard rule. This is fraud detection that adapts to the person, not a one-size-fits-all limit.

auditpulse.io / risk-intelligence
Risk Intelligence
WatchlistAll signalsView alertsConfigure↻ Refresh
Risk Intelligence (ML)Behavioural risk layerAML & ComplianceRule-based AML monitoring

Risk Intelligence

ML layer learns normal patterns and flags unusual activity.

ML pipeline is off for this tenant…Learning member baselines from live channels…

Summary

Open ML signals, watchlist pressure, and channel coverage at a glance.

InactiveActive
Open ML signals
01
Watchlist count
01
Avg 7-day delta+12%
Live channels3
BEHAVIOURAL PROFILEMEM-154031Normal zone · learned baselineOut of character
Channel coverage
CoreMobileSecurityLendingInternetAgency

Recent signals

Latest open behavioural signals — work these in the Signals inbox.

Open watchlist →
ELEVATEDMEM-154031 · Out-of-pattern transferJust now
auditpulse.io / risk-intelligence / signals
Risk Intelligence↻ Refresh
Risk Intelligence (ML)Behavioural risk layerAML & ComplianceRule-based AML monitoring

Behavioural signals

ML-detected unusual activity. Work signals here — not rule-based fraud alerts.

Signal inbox

0 open1 open

These are behavioural risk hits — not the same as rule-based fraud alerts.

AllOpenAcknowledgedClosed
All severitiesHighMediumLow

No signals found

Signals appear when behaviour deviates from learned norms.

HIGHOpen

Velocity anomaly · MEM-154031

Transfer pattern deviates from learned baseline — 3× typical velocity in 30 min.

Behavioural MLJust now
Behavioural Signals

The moment behaviour breaks pattern, it lands here.

Every behavioural risk hit is captured in a dedicated signal inbox — separate from rule-based fraud alerts — and filterable by severity, type, and review status. Auditors work through open signals, acknowledge what needs a closer look, and close what's explained. Nothing gets buried in a general alerts feed.

Risk Watchlist

Not one bad transaction. A pattern building over time.

AuditPulse tracks each member's risk trajectory, not just isolated events — surfacing those with sustained, rising risk before a single transaction ever crosses a hard threshold. Auditors get a prioritised watchlist, a plain-English weekly brief, and a full profile behind every entry, so investigations start with context already attached.

auditpulse.io / risk-intelligence / watchlist
Risk Intelligence
✦ Generate weekly brief

Risk watchlist

Elevated entities · sustained behavioural risk over time

Watchlist is empty

MEM-154031Elevated+34% · 7d

18-day heat cluster · trajectory rising

Weekly brief · generated

MEM-154031 — risk rose steadily over three weeks…

auditpulse.io / aml / alerts
Risk Intelligence
Risk Intelligence (ML)AML & Compliance

AML Alerts

↻ Refresh

Filtered AML compliance alerts — separate from general fraud feeds

AllOpenAcknowledgedClosed
Round-tripping (AML_ROUND_TRIPPING) ▾Apply
General alert noiseHigh value transferFailed login ×5Velocity anomalyStatement downloadsAfter-hours adminPaybill patternDuplicate txnZero-charge flag
AMLCOMPLIANCE GATE
!

No alerts found

Round-tripping (AML)HIGH

AML_ROUND_TRIPPING · MEM-154031

Circular transfer pattern detected — funds returned within 24h window

Open
Documented resolution on record
AML Alerts

One inbox, dedicated to the pattern regulators ask about most.

Round-tripping and related AML rules get their own dedicated alert inbox — kept separate from general fraud and behavioural alerts so compliance reviews never get lost in the noise. Filter by rule and status, and work each alert through to a documented resolution.

Configurable by Design

Every SACCO's risk appetite is different. So is the setting.

Sensitivity, learning periods, and AI-generated narratives are tuned per SACCO — not fixed platform-wide. Rules keep running independently regardless of these settings, so switching behavioural scoring on or off never leaves core fraud detection exposed.

auditpulse.io / risk-intelligence / settings
Risk Intelligence

Risk Intelligence settings

Tune the behavioural ML layer for this tenant.

Risk appetiteInactiveCalibratingActive
ConservativeBalancedAggressive
Core fraud rules · running independently
Tenant configuration
Sensitivity
LowMediumHigh
Minimum history (days)
143060
Signal threshold
Tenant-specific · saved