SEON

Rules based and machine learning decisioning are offered side by side, on a foundation of more than 900 first party signals gathered by querying identifiers against online platforms, breach sources and reputation databases. An enrichment and rules product would keep working without the models.

Customers author risk policies through a scoring engine and a drag and drop workflow builder, starting from shipped default rule sets, with case management and centralized audit trails. Inspectable models show an analyst the signals behind a decision rather than a bare score.

Model updates appear in a running public changelog, and an AI disclaimer updated May 2026 sits on the legal and security hub. Documentation packages, validation summaries and drift monitoring cadence are absent.

More than 5,000 businesses are served, with claims of up to 99 percent fewer multi accounting attempts and 75 percent less manual review time. Testimonials carry only first names and job titles, and a claim of more than $300 billion in prevented losses has no method.

A separate model instance is trained for each customer rather than pooling data into one model, and data from end user checks is processed for one year unless the customer instructs otherwise. Model hosting terms, supplier retention and lifecycle detail are not given.

An annex to the July 27, 2026 privacy notice covers people checked through digital footprint enrichment. It names SEON as processor and describes public, social media and database sources. A subprocessor list is neither named nor linked.

ISO 27001 certification and a SOC 2 Type II report are named, with a trust center on Vanta linked from the legal hub. Certificate dates, audit periods and scope are missing.

Tooling is built around the money laundering reporting officer, a formally appointed role under United Kingdom and European regimes, with sanctions and politically exposed person screening through a dedicated entity API. Obligations stay with the institution, and no financial license is claimed.

Thin digital footprints read as fake identities when scoring weighs online presence, email domain age and breach appearances, which disadvantages older people, recent arrivals and anyone who avoids social media. Demographic error analysis and fairness testing are unpublished.

Checked people may object to inclusion through the customer or SEON, a door that ends at objection rather than a corrected score. Accuracy guarantees and error rates are absent from what is published.

Enrichment sources are typed as social platforms, breach sources and reputation databases, and providers are described by category. Specific platforms, data brokers and cloud providers are unnamed, and no subprocessor list is linked.

Standalone email, phone and address APIs sit beside a modular fraud API returning enrichment, rule evaluation and score in one response, with SDKs for JavaScript, iOS and Android. Which core banking or payment platforms it plugs into is not stated.

Data moves to the United Kingdom and various third countries under standard contractual clauses and the EU US Data Privacy Framework, a transfer mechanism rather than a residency commitment. Hosting provider, region and in country options are unnamed.

Starter costs $699 a month for 2,500 fraud checks, 10 users and 50 custom rules, while Premium is custom priced and holds the anti money laundering tooling a bank would buy. The rate for checks beyond the Starter allowance is not published.

Fintech, digital banking and crypto are covered with screening, transaction monitoring and case management alongside fraud tooling. Gaming, ecommerce and marketplaces carry real weight, and credit unions, wealth and insurance get no material.