Loci Fraud API Products

Explainable Intelligence for Modern Fraud Defense

Loci is a real-time fraud and risk intelligence platform built for financial institutions. We combine transaction monitoring, behavioral analytics, network intelligence, and sanctions screening into a single, explainable system that delivers sub-200ms decisions.

Why Loci?

Traditional fraud systems force a painful choice: static rules that can't adapt, or black-box AI that can't explain itself. Loci offers a third way.

The Loci Platform

Core Products

The Loci Flywheel

When used together, Loci's products create a self-reinforcing intelligence loop:

How it works:

1. Call centrality first — Get the entity's network position before evaluating
2. Submit transaction — MADIE evaluates with network context
3. Graph updates — Transaction stored, entity relationships learned
4. Better next score — Next centrality call has more data

This flywheel means your fraud detection improves automatically with every transaction—no retraining, no manual tuning.

┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│  Your Systems   │────▶│   Loci API      │────▶│    Decision     │
│  (CBS, Mobile,  │     │   Gateway       │     │    + Reasons    │
│   Card Rails)   │     │                 │     │                 │
└─────────────────┘     └────────┬────────┘     └─────────────────┘
                                 │
                    ┌────────────┼────────────┐
                    ▼            ▼            ▼
              ┌──────────┐ ┌──────────┐ ┌──────────┐
              │  MADIE   │ │ Access   │ │  Graph   │
              │  Engine  │ │  Gate    │ │ Analytics│
              └──────────┘ └──────────┘ └──────────┘

1. Ingest

Send transactions, events, or entity data via REST API. Loci accepts data from any source—core banking, mobile apps, card switches, or third-party systems.

2. Evaluate

The MADIE engine evaluates your data against all active rules in parallel. Each rule contributes a weighted score based on the signals it detects.

3. Decide

Loci returns a decision (approve, review, or decline) with a risk score and full explanation of which rules fired and why.

4. Act

Integrate the decision into your workflow. Block high-risk transactions, route medium-risk to review queues, and let low-risk flow through.

Key Capabilities

Fraud Language Models (FLM)

FLM is our domain-specific language for expressing fraud detection logic. Unlike rigid if-then rules, FLMs use weighted evaluations that accumulate evidence across multiple signals.

# Example: Detect potential structuring
name: structuring_detection
description: Flag transactions that may be splitting large sums

evaluations:
  - name: velocity_check
    weight: 4
    # Multiple transfers in short window
    
  - name: amount_similarity
    weight: 5
    # Transfer amounts suspiciously uniform
    
  - name: aggregate_threshold
    weight: 3
    # Total approaches reporting limits

threshold: 0.6

Network Intelligence

Loci builds a real-time graph of transaction relationships. This enables detection of:

• Mule networks: Accounts receiving and dispersing funds rapidly
• Fraud rings: Coordinated activity across multiple accounts
• Hub entities: Central nodes connecting suspicious actors

Adaptive Calibration

Loci automatically calibrates scoring thresholds based on your actual fraud data—not just flagged transactions. This prevents the "death spiral" where false positives contaminate your baselines.

Integration

REST API

curl -X POST https://api.runloci.com/v1/transaction/{orgId} \
  -H "x-org-id: your_org_id" \
  -H "x-api-key: your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "transaction_id": "TXN-12345",
    "entity_id": "CUST-001",
    "amount": 50000,
    "currency": "NGN",
    "transaction_type": "TRANSFER",
    "beneficiary_account_number": "1234567890"
  }'

Response

{
  "success": true,
  "decision": "review",
  "fraud_score": 67,
  "explanations": [
    {
      "rule": "velocity_breach",
      "weight": 4,
      "description": "8 transactions in last hour exceeds threshold of 5"
    },
    {
      "rule": "new_beneficiary",
      "weight": 2,
      "description": "First transfer to this beneficiary"
    }
  ]
}

Compliance & Security

Regulatory Alignment

Loci is designed to meet the explainability requirements of:

• EU AI Act: Full decision lineage and human oversight
• US Model Risk Management (SR 11-7): Auditable, documented decisions
• African Central Bank Guidelines: Real-time monitoring with clear escalation

Data Security

• End-to-end encryption (TLS 1.3)
• Data residency options (EU, US, Africa)
• GDPR/NDPR-compliant data handling

Getting Started

1. Get API Credentials

Contact [email protected] to set up your sandbox environment.

2. Explore the Sandbox

Test API endpoints with sample data. No production impact.

3. Configure Rules

Use Loci Studio to create or customize fraud rules for your use case.

4. Shadow Mode

Deploy rules in shadow mode to measure performance before going live.

5. Go Live

Switch to production with confidence.

Choosing the Right Product

"I need to detect fraudulent transactions" → Start with Transaction Monitoring

"I need to prevent account takeover" → Start with AccessGate

"I need sanctions screening with relationship intelligence" → Start with AccessGate AML

"I need comprehensive protection" → Use all three. They share intelligence automatically.

Integration Architecture