Stop fraud before
a transaction is executed.
Detection · Analysis · Reports
Caspix integrates into your authorization step — before any transaction is approved. A single API call returns a decision in under 200ms, verified against live Safaricom and Airtel Kenya network intelligence. Every alert that comes through has a complete analysis workflow behind it.
Built for institutions across
Built by practitioners with direct fraud operations experience at Kenyan financial institutions.
KES 87,500 · Mobile Transfer
0712 *** 789 → New beneficiary
Contributing signals
Fraud is quicker than the available controls
Losses often surface long after funds have already moved.
Average detection time without automated monitoring
Delayed detection increases financial and evidentiary loss.
Losses associated with internal fraud and collusion
Access-based fraud requires behavior-aware monitoring of staff activity.
False-positive rate in untuned systems
Alert noise burns analyst time and masks genuine high-risk activity.
Based on ACFE Global Fraud Study data and sector estimates from East African financial institutions.
Where Caspix invests differently
Block before funds move
Most fraud systems generate an alert after a transaction has settled and the money is already gone. Caspix integrates into your authorization step before the transaction is approved. A single API call returns a decision in under 200 milliseconds. If Caspix is unreachable for any reason, the transaction proceeds normally.
Verified against the live network
When a transaction involves a mobile-linked account, Caspix queries SIM swap history and subscriber status directly with Safaricom and Airtel Kenya and not batch check against a static database before returning a score. A live query, in the same pipeline pass that evaluates every transaction so that a SIM swap confirmed an hour ago is caught before the funds are spent.
The full analysis in plain English
An analyst types a question about a transaction, a phone number, a pattern across accounts, or an open case and receives a structured analysis drawn from live platform data. SIM swap history, linked accounts, recent alerts, and risk signal attribution all in seconds, without building a query or navigating multiple dashboards. Every frontline analyst has the same capability as the most technically skilled member of your team.
› Show me what is unusual about 0712 456 789 in the last 10 days
3 alerts. Two transactions between 01:20–03:47 EAT — outside the account's 14-month activity window. Safaricom SIM swap confirmed 9 days ago...
Confirmed directly with Safaricom and Airtel Kenya at the moment of scoring
Every mobile transaction triggers a direct network query before a score is returned. This is not a cached record from a third-party database. The network's own answer, in real time.
Eleven capabilities in one analysis environment
Pre-authorization scoring, real-time detection, autonomous analysis agents, plain-English Copilot, and reporting connected in a single workflow.
Pre-Authorization Scoring
Query Caspix before a transaction is authorized — ALLOW, CHALLENGE, or BLOCK in under 200ms. Fail-open by design: unavailability never generates a false block.
Transaction Monitoring
Every transaction is analyzed in real time against threshold, velocity, and behavioral controls — across every channel simultaneously.
AI Risk Scoring Engine
Composite 0–100 score from ten concurrent intelligence modules, a five-model weighted ML ensemble, and a live fraud network graph — ALLOW, CHALLENGE, or BLOCK in under 200 milliseconds.
Rule Engine
Create, test, and deploy fraud rules without waiting for engineering cycles.
Alert Analysis Copilot
Analysts describe what they need in plain English and receive analyst-grade responses from live platform data — no SQL, no dashboard navigation.
Autonomous Analysis Agents
Six specialist agents live in production: analysis briefs, rule optimization, insider threat detection, SAR drafting, executive intelligence reporting, and alert triage — all pulling live platform data.
Alert Triage & Case Management
Structured workflows from alert triage through closure with full analyst attribution and SLA tracking.
Entity Intelligence Registry
Known fraudulent accounts, devices, SIMs, and IDs checked on every transaction. One confirmed fraud protects every institution on the platform.
SIEM Integration
Structured event output for Splunk, Wazuh, Elastic, and QRadar.
Regulatory Reporting
Submission-ready case and escalation reporting for CBK and supervisory requirements.
Audit Trail and Forensics
Immutable activity records for formal review, legal process, and regulator examination.
The analysis interface analysts actually use
Frontline analysts should not need database skills to query a suspicious account. They describe what they need, and the Copilot assembles a structured analysis response from live transaction, alert, case, and account data with full source attribution on every claim it makes.
- Ask about any account, phone number, device, or transaction by name
- Follow-up questions carry the full conversation context — no need to restate the analysis
- Every response cites specific data points: account IDs, amounts, dates, signal attributions
- East African fraud context built in — SIM swap, mobile money, agent banking, insider collusion
›Show me what is unusual about account 284710 in the last 10 days
›Draft a Suspicious Activity Report for this account
The journey from a transaction to a closed case
A structured workflow that shortens analysis time and strengthens audit posture.
Ingest
Transactions enter through API or batch channels
Score
Risk scored across behavioral signals, trained classifiers, and fraud network graph analysis
Alert
High-risk activity is routed for analyst review
Investigate
Evidence captured, linked, and escalated when needed
Report
Cases are closed with full reporting and audit trace
Ingest
Transactions enter through API or batch channels
Score
Risk scored across behavioral signals, trained classifiers, and fraud network graph analysis
Alert
High-risk activity is routed for analyst review
Investigate
Evidence captured, linked, and escalated when needed
Report
Cases are closed with full reporting and audit trace
Built around real fraud typologies
Fraud in a Tier-1 bank may not be exactly the same as fraud in a Tier-3 bank or an MFI. Caspix is tuned to institution type and East African operating realities.
Explore Solutions by InstitutionTier-1 Banks
Insider risk, high-volume alert management, mobile banking integration fraud
Tier-2 Banks
SIM swap, agent banking abuse, cross-channel fraud
Tier-3 Banks
Account takeover, mobile banking fraud, insider privilege abuse
Microfinance (Kenya)
Mobile banking fraud, social engineering, internal fraud and collusion
Fraud Incident & Intelligence Tracker (FIIT)
Caspix flags fraud. FIIT is where your team takes it from alert to recovered funds — incident tracking, partner validations, hold requests, chargeback management, and recovery analytics. Included with every Caspix subscription.
Security and compliance are built-in controls
RBAC, encrypted audit logs, secrets management, and deployment patterns aligned to CBK regulatory expectations.
Map the platform to your highest-risk fraud scenarios.
Book a structured demo and we will map the platform to your highest-risk transaction scenarios.