SoniqPay
Platform
Platform overviewGateway & VaultOrchestration & Smart RoutingAI Fraud PreventionGuaranteed TransactionsLost Transaction RecoveryConsolidated Reporting
Company
IndustriesDevelopersCompanyContactLegal notices
Platform / AI Fraud Prevention

Risk decisions that understand your vertical, not retail averages.

Most fraud tools are calibrated on physical-goods e-commerce and then pointed at everything else. SoniqPay scores each transaction against models trained on the behavior of the vertical it came from — deposits, top-ups, micro-transactions, and payouts behave nothing like a shipped order.

Decision point
Pre-authorization
Signal classes
Device, behavior, instrument, network
Score latency
Inline with the auth call
Rule control
Yours, per merchant
How it works

AI Fraud Prevention, end to end.

Four stages, each observable in the dashboard and addressable over the API.

Step 1

Collect

Device fingerprint, session behavior, instrument history, and velocity across your whole portfolio.

Step 2

Score

An ensemble model returns a risk score and the specific features that drove it.

Step 3

Decide

Approve, challenge with 3DS, route to a stricter MID, or block — a policy you set, not one we impose.

Step 4

Review

Every decision is explainable in the dashboard, with the option to label outcomes and retrain.

Where the model sits

Scoring happens before the authorization, not after the loss.

A decision that arrives after settlement is a report, not a control. SoniqPay scores inline with the authorization call, so the outcome can change the route, trigger an authentication step, or stop the transaction while stopping it still costs nothing.

Inline with authThe score is produced in the same call path, without a second round trip from your servers.
Feeds the routerA high score can be routed to a stricter MID rather than declined outright.
Gate for the guaranteeThe same score determines which transactions qualify for chargeback protection.
Signal classWhat it looks at
DeviceFingerprint stability, emulator and automation detection, IP reputation, proxy and VPN posture
BehavioralSession shape, input cadence, navigation path, time-to-submit
InstrumentBIN intelligence, issuer response history, token age, prior dispute record
VelocityAttempts per card, device, email, and account across your entire portfolio
ContextualAmount relative to customer history, geography mismatch, time-of-day pattern
Capabilities

What you get.

Everything below ships as part of ai fraud prevention — no separate module, no separate contract.

Explainable scoring

Each score comes with the contributing features ranked, so an analyst can defend the decision.

Velocity across the portfolio

Catch the card, device, or email cycling between your merchants, not just within one.

Behavioral signals

Typing cadence, session shape, and navigation patterns that separate a person from a script.

Policy over verdict

The model produces a score; your policy decides what happens at each threshold.

Challenge instead of block

Step up to 3DS or a stricter route rather than throwing away a marginal good customer.

Feedback loop

Chargeback and dispute outcomes flow back automatically to retrain against your real losses.

Bring your own MID. We will bring everything else.

Early access is opening to platforms running real volume. Tell us your stack and we will tell you plainly where the technology moves your numbers.