vcurd is India's leading real-time UPI fraud detection and rule management platform — built ground-up for India's digital payment ecosystem. Detect, flag, and block fraudulent UPI transactions in under 50ms with full SHAP explainability, RBI audit trails, and AI-generated rule suggestions.
Built for organisations empanelled with
vcurd integrates with top AI platforms to deliver intelligent fraud detection, automated pattern analysis, and smart rule generation — fully configurable.
Advanced reasoning for fraud pattern analysis and intelligent rule generation with guaranteed structured JSON output. Best for speed and accuracy.
Constitutional AI with exceptional instruction-following for nuanced compliance reasoning, risk narrative generation, and regulatory-ready audit summaries.
Multi-modal intelligence for analysing transaction patterns across structured data, logs, and unstructured fraud signals with deep contextual understanding.
Run open-source models (Llama 3, Phi-3, Mistral) entirely on your own infrastructure. Zero data leaves your servers — ideal for air-gapped environments.
vcurd's proprietary fraud detection model — custom-trained exclusively on Indian UPI transaction patterns. Unlike generic AI models, it understands the nuances of VPA structures, Indian merchant categories, seasonal velocity patterns, and regional fraud signatures. Delivers unmatched precision for Indian financial institutions with full SHAP explainability on every decision.
16 integrated modules working together to give you complete visibility and control over every risk vector in India's digital payment ecosystem.
Every transaction is scored by a trained fraud detection model in under 50ms. Each decision comes with a probability score and SHAP-based feature attribution — no black boxes.
Compliance and risk teams can create, toggle, and tune detection rules in real time — no deployments needed. Rules evaluate before ML scoring for maximum speed.
BIN lookup and card network analysis to detect international card misuse, prepaid card fraud, and high-risk card categories at transaction time.
NetworkX directed graph analysis of VPA-to-VPA flows assigns each account a 0–100 mule score based on in-degree, out-degree, and fan-out patterns.
PAN-level loan-stacking detection tracks hard-pull frequency across the 72-hour window and triggers an automatic block when velocity exceeds configured thresholds.
Real-time LTV calculation against gold valuations with RBI 75% threshold alerts. Every breach is logged with full history for regulatory review.
Daily balance versus expected reconciliation with 15% variance breach detection. Consecutive breach alerts escalate automatically for compliance review.
Correlates SIM-swap events with Account Aggregator consent patterns. Detects multi-FIU burst requests that indicate credential takeover or synthetic identity fraud.
District-level fraud volume plotted on an India map with MHA high-risk zones highlighted. Bubble size scales with fraud count for instant visual triage.
Directed VPA-to-VPA flow graph exposes mule rings, fan-out money movement, and layered fraud networks invisible to transaction-level analysis.
A multi-agent AI pipeline runs autonomously in the background — generating fraud pattern reports, anomaly summaries, and risk briefings without manual intervention.
Monitor transaction volume, fraud rates, triggered rules, and velocity trends in real time via WebSocket. Drill into any transaction to see exactly why it was flagged.
Every rule evaluation, model decision, and manual override is logged with timestamps, user attribution, and full context — ready for regulatory review at any time.
Countdown timers to RBI reporting deadlines, one-click FMR JSON export, and pre-formatted compliance packages for regulatory submission.
Upload CSV files for bulk ML scoring with asynchronous job tracking. Ideal for end-of-day reconciliation, historical re-scoring, and compliance audits.
Centralised lookup for mule scores, bureau velocity, suspicious VPA registry, and escrow history — all in one unified dashboard for operations teams.
Every risk vector in India's digital payment ecosystem — covered.
A five-layer defence that combines deterministic rules, probabilistic ML, and enriched signals — giving you the speed of heuristics and the intelligence of AI.
Every UPI transaction hits the vcurd API. Metadata — amount, merchant category, device, time, velocity — is extracted and enriched in real time.
The transaction is evaluated against all active rules in priority order. A BLOCK rule stops it immediately; FLAG / DELAY / 2FA rules apply their action and continue.
The trained XGBoost pipeline scores the transaction, producing a fraud probability with SHAP explanations. The final verdict — APPROVE / HOLD / BLOCK — is returned with full reasoning.
In parallel: mule score lookup, bureau velocity check, fan-out graph analysis, and consent/AA anomaly detection. All signals are attached to the transaction record.
An alert is created, the RBI compliance countdown is updated, the compliance log is written, and a WebSocket broadcast is sent to all dashboard clients in real time.
{
"transaction_id": "UPI240519182344",
"amount": 49500,
"verdict": "BLOCK",
"rule_triggered": "high_value_new_beneficiary",
"fraud_probability": 0.94,
"mule_score": 87,
"bureau_velocity_72h": 4,
"location_district": "Jamtara",
"risk_level": "HIGH",
"top_features": [
{ "feature": "is_new_beneficiary", "impact": +0.38 },
{ "feature": "mule_score", "impact": +0.31 },
{ "feature": "bureau_velocity", "impact": +0.19 }
],
"latency_ms": 43
}Role-specific views and workflows designed around how your analysts, compliance officers, engineers, and ops leads actually work.
Real-time transaction feed with live WebSocket updates, mule score lookups, network graph drill-down, and a suspicious VPA management console — everything needed to investigate and action fraud in real time.
RBI reporting countdown timers, one-click FMR export, gold LTV breach alerts, escrow variance monitoring, and a full immutable audit trail ready for regulatory submission at any time.
No-code rule engine with AI-powered rule suggestions, CSV batch processing, API-first architecture, and configurable ML thresholds — build and iterate without raising a ticket.
Live analytics dashboard, geographic fraud heatmap, bureau velocity alerts, and a consent/AA anomaly dashboard — complete operational visibility across every risk dimension.
Not a generic fraud tool adapted for India — built ground-up for UPI, with Indian payment patterns, regulatory requirements, and compliance workflows in mind.
Designed specifically for UPI transaction patterns, VPA-based blacklisting, and Indian banking infrastructure. Not a generic fraud engine.
SHAP-powered explanations on every ML decision. Your compliance team can justify any block or flag to regulators and customers.
Risk teams change rules without raising tickets. Engineers stay focused on product — not on updating fraud thresholds.
Built on FastAPI + PostgreSQL + Redis with async processing. Handles high transaction volumes without degrading detection quality.
Every decision is logged. Every rule change is tracked. Every model score is stored. One-click export for regulatory submissions.
API-first design. Integrate seamlessly into your existing security stack or offer as a managed service to your financial clients.
| Feature | vcurd | Legacy Rule Systems | Generic ML Platforms |
|---|---|---|---|
| Real-time scoring (<50ms) | ✓ | ✗ | ✓ |
| No-code rule management | ✓ | ✗ | ✗ |
| AI rule suggestions | ✓ | ✗ | ✗ |
| SHAP explainability | ✓ | ✗ | Partial |
| UPI-native features | ✓ | ✗ | ✗ |
| Full audit trail | ✓ | Partial | ✗ |
| Multi-platform AI | ✓ | ✗ | ✗ |
| Swarm AI analytics | ✓ | ✗ | ✗ |
| Card BIN intelligence | ✓ | ✗ | ✗ |
| Mule account detection | ✓ | ✗ | ✗ |
| Geographic heatmap | ✓ | ✗ | ✗ |
| Bureau velocity monitoring | ✓ | ✗ | ✗ |
| Gold LTV / Escrow monitoring | ✓ | ✗ | ✗ |
| Consent/AA anomaly detection | ✓ | ✗ | ✗ |
| RBI compliance reporting | ✓ | Partial | ✗ |
| Network graph analysis | ✓ | ✗ | ✗ |
| Real-time WebSocket feed | ✓ | ✗ | ✗ |
vcurd is used by banks, fintechs, payment aggregators, NBFCs, and MFIs across India — each with purpose-built modules for their specific fraud and compliance challenges.
Real-time UPI blocking, mule network detection, FMR reporting, and full RBI examination readiness.
API-first integration, no-code rule management, AI rule suggestions, and multi-LLM fraud intelligence.
Cross-merchant fraud pattern detection, escrow compliance monitoring, and suspicious VPA registry.
Advance fee fraud, gold LTV monitoring, JLG device sharing, bureau velocity, and disbursal controls.
Common questions from banks, fintechs, and compliance teams evaluating vcurd.
Talk to us about deploying vcurd for your financial institution or cybersecurity practice.
Whether you're a CERT-In empanelled organisation, a bank, a fintech, or a payment aggregator — reach out and we'll show you what vcurd can do for your clients.
partners@vcurd.com
Send an email →Gurgaon, India
Available for in-person meetingsAPI integration and partnership arrangements for CERT-In empanelled organisations.
Get in touch →We respond to all enquiries within 24 hours on business days.
Mon – Fri, 9 AM – 6 PM IST