Anti Money Laundering & Risk Intelligence
Detect risk before it becomes loss.
Machine learning and rule-based monitoring — built for SACCO compliance and audit teams.
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.
Risk Intelligence
ML layer learns normal patterns and flags unusual activity.
Summary
Open ML signals, watchlist pressure, and channel coverage at a glance.
Recent signals
Latest open behavioural signals — work these in the Signals inbox.
Behavioural signals
ML-detected unusual activity. Work signals here — not rule-based fraud alerts.
Signal inbox
0 open1 openThese are behavioural risk hits — not the same as rule-based fraud alerts.
No signals found
Signals appear when behaviour deviates from learned norms.
Velocity anomaly · MEM-154031
Transfer pattern deviates from learned baseline — 3× typical velocity in 30 min.
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.
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.
Risk watchlist
Elevated entities · sustained behavioural risk over time
Watchlist is empty
18-day heat cluster · trajectory rising
MEM-154031 — risk rose steadily over three weeks…
AML Alerts
↻ RefreshFiltered AML compliance alerts — separate from general fraud feeds
No alerts found
AML_ROUND_TRIPPING · MEM-154031
Circular transfer pattern detected — funds returned within 24h window
OpenOne 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.
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.
Risk Intelligence settings
Tune the behavioural ML layer for this tenant.