Enterprise-grade fraud prevention for banks, fintechs, and payment platforms. Powered by 5 AI models operating in parallel — decisions delivered in under 50ms with 99.7% accuracy.
A single API call triggers an ensemble of 5 parallel AI models. Every factor — velocity, geography, device, behavior, network — analyzed simultaneously before a decision is returned.
Your backend sends a single POST request with transaction metadata: amount, user ID, device, IP address, and location context.
VelocityAnalyzer, GeoSentinel, DeviceGuard, BehavioralIQ, and NetworkGraph each score the transaction independently against 10B+ historical data points.
A unified risk score (0.0–1.0) and recommendation — APPROVE, REVIEW, or BLOCK — is returned alongside detailed factor analysis and an audit-ready trace.
Each model is trained on a distinct fraud signal domain. Their outputs are fused by the Ensemble Layer — giving you defense-in-depth against every attack vector.
Detects abnormal transaction frequency and timing patterns. Identifies card-testing attacks, rapid account takeovers, and burst-fraud sequences before they escalate.
Cross-references transaction location against a user's historical geography. Flags impossible travel, proxy usage, and high-risk jurisdiction patterns with per-user baseline.
Analyzes device fingerprint, browser entropy, and hardware signatures. Detects emulators, headless browsers, device spoofing, and SIM-swap attacks in real time.
Builds a rolling behavioral fingerprint per user: spending patterns, session timing, interaction cadences, and contextual norms. Any statistical deviation triggers investigation.
Maps relationships between users, merchants, devices, and IP addresses. Uncovers fraud rings, money mule networks, and coordinated synthetic identity schemes.
Combines all five model outputs using a meta-learner trained on 10B+ transactions. Applies your custom rules on top. Final risk score and recommendation — single, unified, explainable.
FraudGuard processes millions of live transactions daily. Every single one is scored in real time — no batch processing, no delays, no exceptions.
import FraudGuard from '@vacalion/fraudguard-sdk'; // Initialize once — reuse across your app const fraud = new FraudGuard({ apiKey: process.env.VACALION_API_KEY }); // Analyze before settling any transaction const result = await fraud.analyze({ transactionId: txn.id, amount: txn.amountCents, currency: "USD", userId: user.id, ipAddress: req.ip, deviceId: req.headers['x-device-id'] }); // Act on the verdict in milliseconds if (result.recommendation === "BLOCK") { await rejectTransaction(txn.id); return { error: "Transaction blocked" }; } await settleTransaction(txn.id);
Official SDKs for Node.js, Python, Ruby, and Java. A single API call with a handful of fields is all you need to protect every transaction.
Every transaction receives a composite risk score between 0 and 1 in under 50 milliseconds. Three-tier verdicts — APPROVE, REVIEW, BLOCK — with configurable thresholds per merchant or product segment.
Risk score, recommendation, and factor analysis returned before the transaction settles.
APPROVE, REVIEW, or BLOCK — plus configurable score thresholds per merchant or product line.
Each response includes the model signals that drove the score — enabling manual review decisions you can defend to regulators.
BehavioralIQ builds a continuously updated fingerprint of each user's normal behavior — spending patterns, device habits, session rhythms. Any statistical deviation triggers elevated review, silently, without friction for legitimate customers.
A unique statistical model per user — trained automatically from their transaction history from day one.
99.99% of genuine transactions pass in milliseconds. Friction only when the signals truly warrant it.
Catches the subtle behavioral shifts that occur immediately after a credential compromise — before financial damage occurs.
Build domain-specific fraud prevention rules that execute alongside the ML models. Set thresholds per merchant category, per country, per user segment. Combine rule triggers with model outputs for maximum precision.
Create and deploy rules through the Developer Dashboard — no code required for business-side configuration.
Deploy new rules instantly — live within 200ms globally. No deployments, no downtime.
Every rule trigger is logged with full context — enabling compliance reporting and model performance analysis.
No setup fees. No hidden costs. Start free, scale as you grow — pay only for what you use.
What's included
Billed annually as $2,292/yr
For scaling fintechs that need real production throughput and full customization capabilities.
Contact SalesEverything in Starter, plus
For banks and large platforms with compliance, data residency, or custom model requirements.
Contact salesEverything in Growth, plus
All plans include SSL encryption, PCI-DSS compliance, and 24/7 fraud monitoring. No long-term contracts on Starter and Growth. Read the API docs →
Start free — no credit card required. Integrate in minutes. Your first 5,000 analyses are on us.