Authorization ServerThe decision engine these fraud rules run inside of, at the moment of authorization.
Solutions
More About Fraud Detection.
Fraud checks that run at the moment of the transaction and not a report you read after the loss.
Explore Payment Solutions
On This Page
Reading time: ~4 Min ReadRule-based velocity and threshold checks run at the moment of authorization, the same decision that approves or declines the transaction, not a separate system reconciling after the fact.
A refueling anomaly layer runs machine learning across every refueling event today - odometer verification, speed and distance plausibility, consumption bounds, and station-distance validation, in pilot with customers now.
Advanced pattern detection: card cloning, impossible travel, station-collusion patterns, is in active development. This page is upfront about what's live today and what's still being built, rather than blurring the two.
What Actually Catches Fraud? A Rule Checking Every Transaction, Or A Report You Read Later?
Multiple rules checking every transaction as they happen. Velocity and threshold checks run at authorization, alongside a refueling anomaly layer that scores odometer, speed, distance, and consumption plausibility on every event - not a fraud report that tells you what already happened.
Velocity & Threshold Rules At Authorization
Rule-based checks: spend velocity, threshold breaches, unusual patterns, evaluated at the same moment a transaction is authorized, not by a separate system after the fact.
Refueling Anomaly Detection, Live Today
Machine learning runs across every refueling event, odometer verification, speed and distance plausibility, consumption bounds, and station-distance validation.
Receipt Image Processing
A receipt image is read automatically, vendor, amount, date, and line items extracted and mapped to the transaction record, without manual data entry.
A Decision Audit Trail
Every approval, decline, and fraud event is logged with a reason, feeding the same evidential trail used for transaction monitoring and reporting.
Human Oversight, Committed By Design
Governance principles apply as new detection capability ships: no personal data used to train shared models without consent, a human in the loop for any adverse automated decision, and explainability documentation per feature.
PCI DSS Compliance
Fraud controls run on a platform holding a current PCI DSS v4.0.1 attestation of compliance as a service provider.
How You Do It
How Does A Fraud Check Actually Run Against A Live Transaction?
Velocity, threshold, and balance rules are evaluated at the same moment authorization is decided, not afterward.
Odometer, speed, distance, and consumption plausibility are checked against the vehicle and driver's own pattern, not a generic threshold.
Approvals, declines, and fraud events are logged with an explicit reason, feeding the same audit trail used for transaction monitoring.
Automated detection informs the decision; an adverse outcome keeps a human in the loop, by design, not as an afterthought.
Post-Hoc Fraud Reports vs. Rule-Based Alerts vs. Reins
How Do They Actually Compare?
Capability
Post-Hoc Fraud Reports
Rule-Based Alerts Only
Real-Time + Anomaly LayerA report that tells you about a fraud pattern after the statement runs means the loss already happened - the report just confirms it.
After the statement or report runs
At authorization, against fixed rules
At authorization, plus per-event anomaly scoring
What's checked
Whatever the report happens to flag
Velocity and threshold only
Velocity, threshold, and refueling-pattern plausibility
Receipt handling
Manual entry or none
Manual entry or none
Automated extraction, mapped to the transaction
Decision transparency
Varies
Varies
Explicit reason on every approval, decline, and fraud event
Where the roadmap stands
Unclear
Unclear
Stated plainly: rule-based and anomaly layer live; advanced pattern detection in development
Reinvent Growth
—
—
Always — the same platform adds wallet, loyalty, and marketing layers as you grow See Solutions
Post-Hoc Fraud Reports
A report that tells you about a fraud pattern after the statement runs means the loss already happened - the report just confirms it.
After the statement or report runs
What's checked
Whatever the report happens to flag
Receipt handling
Manual entry or none
Decision transparency
Varies
Where the roadmap stands
Unclear
Reinvent Growth
—
Rule-Based Alerts Only
A report that tells you about a fraud pattern after the statement runs means the loss already happened - the report just confirms it.
At authorization, against fixed rules
What's checked
Velocity and threshold only
Receipt handling
Manual entry or none
Decision transparency
Varies
Where the roadmap stands
Unclear
Reinvent Growth
—
Reins — Real-Time + Anomaly Layer
A report that tells you about a fraud pattern after the statement runs means the loss already happened - the report just confirms it.
At authorization, plus per-event anomaly scoring
What's checked
Velocity, threshold, and refueling-pattern plausibility
Receipt handling
Automated extraction, mapped to the transaction
Decision transparency
Explicit reason on every approval, decline, and fraud event
Where the roadmap stands
Stated plainly: rule-based and anomaly layer live; advanced pattern detection in development
Reinvent Growth
Always — the same platform adds wallet, loyalty, and marketing layers as you grow See Solutions
FAQ.
Part of the same decision. Velocity and threshold rules are evaluated at authorization, against live account state, not by a separate system reconciling after the fact.
Yes. A refueling anomaly layer runs machine learning across every event: odometer, speed, distance, and consumption plausibility, in pilot with customers today.
No. Receipt image processing extracts vendor, amount, date, and line items automatically and maps them to the transaction record.
Yes. Every approval, decline, and fraud event is logged with an explicit reason, feeding the same audit trail used for transaction monitoring and reporting.
Adverse automated decisions keep a human in the loop by design, with explainability documentation maintained per feature.
That level of pattern detection - card cloning, impossible travel, station-collusion, is in active development, not live today. This page will be updated as it ships.
Yes. Fraud controls run on infrastructure holding a current PCI DSS v4.0.1 attestation of compliance.
Related Solutions.
Real-Time Spend ControlThe spend limits and restrictions these checks are evaluated alongside.
Real-Time AI Insights / BIThe broader analytics layer this anomaly detection feeds into.
