Fraudio

Scores come from a single centralized model, with no rules to author and no per customer configuration, returned in under 100 milliseconds, and the claimed edge over rules and single customer models is itself a modeling claim. Take the model away and no rules engine, case product or separate data asset is left.

Plug and protect is the pitch: no rules, configuration or tuning, with a real time or daily batch score sent to the customer's own system, where the decision is made. Its founders reject black box solutions, yet no reason code, threshold or override is described.

A prospect can send historical transactions, have them scored by the centralized model and receive the detection results free and without obligation, so the model is tested on the buyer's own population before deployment. Precision, recall and false positive rates are unpublished, and no one customer can version the shared model.

Viva Wallet grew transactions sevenfold, raised its risk team's efficiency by 600 percent and earned an eightfold return, by Fraudio's account, with no method or period published. Comparisons claiming up to 30 times better results than single dataset models name none of the comparators.

One payment service provider's transactions train the model that scores a competitor's traffic, because Fraudio connects customers' datasets into one centralized model by design. The terms limit retained query records to monitoring, quality improvement and billing, a list that sets no boundary on training the shared model or on what a departing customer's data keeps doing.

The privacy policy, last updated September 15, 2020, names Fraudio B.V. as controller and Google Analytics as a processor but never reaches the transaction and cardholder data the service handles. Retention periods and a supervisory authority are absent, and the November 2025 terms require only return or destruction of confidential information on completion or request.

ISO/IEC 27001 certification was announced in January 2024, and the mark sits on the homepage. Certification body and scope go unstated, and no payment card attestation is named although the platform ingests card transactions.

License, supervisory relationship and scheme recognition all go unstated, usual for a technology supplier. Anti money laundering and money mule products are sold without citing the directives they serve, though customers face card scheme fraud thresholds the market material describes.

Every customer shares one model, so a pattern learned in one geography or merchant category applies across the network at once rather than staying with the institution whose data produced it. Behind it sit no error rate, calibration statement or fairness testing.

Terms updated November 2025 cap liability at 12 months of fees, warrant nothing about score accuracy beyond workmanlike delivery and require the client to indemnify Fraudio. A cardholder scored as fraudulent by a model trained on other institutions' data gets no notice, explanation or correction route.

Across five products, anti money laundering among them, Fraudio names no data supplier, enrichment provider or watchlist source. Without a subprocessor list, whether any external model sits behind the scores is unknown.

Real time or daily batch data goes in through a single API, which suits an authorization path. Silverflow and Pismo appear as customers, yet no connector, gateway or case management integration is named and no developer documentation is published.

Delivery runs on AWS, shown on the homepage, with no region, residency commitment or on premises option stated. Because the model is centralized, the data must reach one place, and an acquirer whose supervisor expects cardholder data to stay in a jurisdiction would need an answer before signing.

Pay per use on monthly subscriptions with no commitment is the only model, and the pricing page returns a price once a company enters its monthly transaction count. Buyers get a free trial, though no rate or volume tier appears.

Payment service providers, acquirers and issuers are addressed, with firms named across that chain such as Bancontact, PayTabs and Cashflows. Merchant initiated fraud detection serves acquirers watching their own portfolios, though segment material is uneven across the list.