Device Intelligence
Devices tell a story. A new device on an account might be innocent or the start of account takeover. The same device across multiple accounts within minutes is a credential-stuffing attack. Device Intelligence cuts through noise by clustering device fingerprints, detecting patterns, and flagging anomalies.
What you can do
- Identify devices across sessions, networks, and registration attempts
- Detect account takeovers and credential stuffing attacks
- Catch synthetic fraud at registration time
- Flag botnet activity and automated attacks
- Build device reputation scores into your decisioning logic
Machine Learning
Fraud evolves. Attackers test new vectors constantly. Rules written today are already stale tomorrow. Machine Learning models trained on millions of real transactions learn patterns no rule can capture — and adapt to new attack types without manual intervention.
What you can do
- Deploy pre-trained models for card-not-present fraud, account takeover, and friendly fraud
- Tune model sensitivity to match your risk appetite
- Retrain models from your own transaction data for industry-specific patterns
- Monitor model drift and performance in real time
- Blend predictions from multiple models into a single risk score
Case Management
Not every fraud case is clear-cut. Some transactions sit in a grey zone — high risk but not certain. Case Management is a unified queue where analysts review, adjust, and learn. Every decision feeds back into your rules and models.
What you can do
- Route cases by risk tier, payment method, geography, or custom criteria
- Adjust rules and weights from the review dashboard without code
- Retrain models from analyst decisions
- Collaborate on complex cases (team notes, discussion threads)
- Set SLA timers and escalation workflows
Card Payment Fraud
Card-not-present fraud hits your approval rate the hardest. False declines frustrate good customers. Missed fraud becomes chargebacks. Card Payment Fraud detection balances both.
What you can do
- Real-time decisioning for every card transaction
- Reduce false positives through multi-factor risk scoring
- Trigger 3DS challenges only when risk exceeds your threshold
- Detect card testing attacks (small transactions used to validate stolen cards)
- Monitor scheme requirements (Visa VAMP, Mastercard Fraud Monitoring) and flag cases automatically
Merchant-Initiated Fraud
Recurring charges, subscription trials, and credential-on-file transactions are harder to monitor than one-off purchases. Fraudsters exploit delayed chargebacks, trial abuse, and stored-card vulnerabilities in this space. GuardHive detects patterns unique to MIT flows — from trial cycling to dormant-card billing — before chargebacks arrive.
What you can do
- Detect free-trial abuse, credential stacking, and synthetic subscriptions
- Flag unauthorized recurring charges and failed re-authorization patterns
- Catch dormant-card billing abuse and delayed chargeback signals
- Prevent COF (credential-on-file) abuse on stored payment methods
- Block subscription cycling and overlapping trial fraud rings
Technical foundations
Built on Kubernetes with multi-region failover. No on-premise infrastructure.
Every feature — detection, case management, reporting — available via REST API.
Rules are version-controlled. Every decision includes a full audit trail.
EU-based deployments with GDPR-compliant data handling. PSD2, PCI DSS certified.