TrustDecision

Scoring and fraud detection run on models, while device identification, risk personas built from IP, email and phone signals, and the rules a risk team configures would keep working without them. Those models power a core capability inside a product that could still run on rules and device intelligence alone.

ARGUS sets out how a decision runs: a no code rule engine with first match, worst match and weight based execution modes, and full audit logs on every decision. On the retail banking page, cases with a medium risk score go to manual review in case management at a threshold the bank sets.

Risk teams can run real time simulation and batch tests on live data, version a strategy and roll it back, and each action carries an explainable reason tied to audit readiness. Speed is the only performance figure, 20 milliseconds a decision, with no accuracy, false positive rate or validation result and no link to a supervisory model risk standard.

Headline figures are more than 1,000 clients, 120 billion risk events intercepted and about $10 billion a year in losses avoided. Every case study is anonymized, and the one carrying a figure, more than 60 million flash sale abuse attempts blocked, belongs to a collectible toy brand.

A categorical line in the privacy policy says end user data is never shared between clients, which closes off the obvious form of cross client exposure in a device and behavioral risk network. The same policy lists accuracy improvement as a purpose, without saying whether one client's fraud outcomes train models or risk labels served to another.

Coverage extends to end users: the privacy policy, effective September 23, 2025, reaches the browser, device and behavioral data that flows through the SDKs on client sites, and covers transfers under standard contractual clauses and binding corporate rules. Retention runs only as long as necessary, and the processors behind the service are not named.

Two certification logos sit in the footer, but neither is named in text, and no trust center or report is published. Encryption and anonymization get a mention in the privacy policy, which is as far as the security detail goes.

Data protection laws are the only rules named, Singapore's PDPA, the GDPR and Brazil's LGPD among them, and they cover the firm's own processing. Although eKYC and AML are listed among its functions, the platform itself is given no regulatory position, and neither a financial supervisor nor a fraud, AML or credit rule is cited.

Alternative credit scores for thin file and new to credit applicants draw on device and behavioral signals that can stand in for income or social group. Its inclusion case, small initial limits instead of refusal, comes with no published fair lending testing, subgroup results or adverse action guidance.

Correction runs through a Singapore data protection officer, who answers end users within 15 days, but that covers personal data rather than decisions. An applicant declined on an alternative score, or tied to a fraud ring through a shared device, gets no published reason or appeal, and no guarantee or indemnity stands behind the output.

Huawei Cloud and AWS are named as infrastructure partners, while the identity verification providers, hosting providers and affiliates in the privacy policy are referred to only by role. Model and data providers are not identified, and since the policy lets data move among affiliates, those names matter to a bank.

Developers get an open API reference covering device fingerprinting, identity verification and fraud management, with SDKs for Android, iOS and the web and an MCP endpoint for AI agents. Integrations with core banking, origination or case management systems are not listed.

Delivery is described only as cloud native, with no region list, residency commitment or private deployment published. That leaves banks in Singapore, Indonesia or Mexico without a stated home for their data.

Engagement starts with a contact request, and the site carries no plan, price or basis of charge. Charges could follow API calls, decisions or a platform license, and which one applies is never said.

Retail banking, consumer lending and digital payments each get an industry page, and the retail banking page carries its own FAQ and bank services. Solutions run from account protection and application fraud through identity verification to credit risk management, while the client base reaches well into ecommerce, gaming and travel.