Σ π φ

Outsmart Fraud.
Protect Revenue.
At Scale.

Euclid Super Intelligence Engine scores every lead across Identity Risk, Fraud Signal, and Health Outlook Risk — delivering a composite fraud score (0–100) that carriers can act on instantly.

EUCLID SCORE
0
HIGH RISK
IR: 91 FS: 84 HR: 62
AI-Driven Scoring
Explainable Results
PII Protected

Risk Intelligence Engineered for Insurance Carriers

Euclid is the core fraud detection engine within Open Insurance OS. It combines rule-based algorithmic scoring with external API validation and LLM-powered pattern recognition to produce a composite fraud risk score for every lead — processed at scale, in real time.

functions

Composite Scoring

Every lead receives a weighted Euclid Score (0–100) derived from Identity Risk, Fraud Signal, and Health Outlook Risk dimensions.

bolt

Bulk Processing Pipeline

CSV datasets are chunked into 1,000-row segments and distributed across up to 10 parallel edge-function workers. Built for 100K+ rows.

manage_search

External API Validation

ZeroBounce verifies email deliverability and spam-trap exposure. Abstract Phone API identifies VOIP, disposables, and data breach history.

psychology

LLM Fraud Pattern Analysis

Theo Fraud Orchestration LLM performs deep pattern recognition and fraud-ring identification on sampled lead populations.

assessment

Executive Report Generation

AI-generated PDF reports with findings, risk analysis, and carrier recommendations — one click.

shield

PII Protection Mode

Tokenizes and decentralizes names, emails, and phone numbers before any LLM processing. Protected and sensitive modes.

Three Dimensions. One Composite Truth.

Euclid's hybrid scoring model is 70% algorithmic, 10% email API, 10% phone API, and 10% AI pattern analysis. Every score is explainable — top risk factors are surfaced with weights and human-readable reasoning.

IR

Identity Risk

  • Email/phone validation
  • Duplicate detection
  • Product-age mismatches
80% of IRC
FS

Fraud Signal

  • Phone/email reuse across identities
  • Submission velocity
  • VOIP detection
  • Address & residency verification
20% of IRC
HR

Health Outlook Risk

  • Age bands
  • BMI
  • Chronic conditions
  • Expected Remaining Years (ERY)
10% of Euclid Score
IRC = 0.2 × FS + 0.8 × IR
Euclid Score = 0.9 × IRC + 0.1 × HR
LOW (0–39) MEDIUM (40–69) HIGH (70–100)

Everything Carriers Need. Nothing They Don't.

LEAD VALIDATION — #EU-00412
Email Deliverability ✓ VALID
Email Domain Reputation ✓ CLEAN
Phone Line Type ✗ VOIP
Disposable Number ✗ FLAGGED
Address Verification ✓ MATCH
Residency Verification ⚠ MISMATCH
Duplicate Identity ✗ 3 HITS
[ Multi-Point Validation ]

Every Data Point, Cross-Referenced

Euclid validates email deliverability, spam-trap exposure, phone line types, VOIP and disposable number detection, physical address match, and residency verification — all in a single pass per lead.

[ Agent Intelligence ]

Agent Performance Ratio. Commission Control.

Euclid calculates an Agent Performance Ratio (APR) using z-score normalization across six performance metrics with volume shrinkage. Carriers can proactively flag or withhold commissions from agents whose portfolio patterns indicate fraud exposure.

AGENT PROFILE — AGT-2291
APR Score 0.31 — BELOW THRESHOLD
Leads Submitted 847
High Risk Rate 68%
Submission Velocity ↑ ELEVATED
Email Reuse Flags 12
Phone Reuse Flags 9
TheoAI Copilot
"Summarize high-risk agents in this dataset"
3 agents flagged as HIGH RISK
AGT-2291: APR 0.31, 68% high-risk leads
AGT-1847: Phone reuse across 14 identities
AGT-0093: Submission velocity 4.2× avg
Report generated — 847 leads analyzed
[ AI Intelligence ]

Ask Anything. Get Answers Instantly.

TheoAI Copilot lets analysts query datasets in natural language — citing specific leads and agents, surfacing fraud patterns, and triggering AI-generated executive PDF reports with risk findings and recommendations.

Built for the Volume of Real Insurance Operations

0–0
Composite Score Range
Parallel Worker Threads
0
Rows Per Processing Chunk
0
Scoring Dimensions Per Lead

Hybrid Scoring Breakdown

Algorithmic70%
Email API10%
Phone API10%
AI Pattern10%

Validation Sources

ZeroBounce
Email deliverability, spam-trap, MX validation, quality scoring
Abstract Phone API
Line type, carrier info, data breach history
Google Maps
Geographic visualization, risk heatmaps
Theo LLM
Fraud-ring pattern recognition

Terms & Definitions

All fraud detection terms, acronyms, and concepts used throughout the Euclid Super Intelligence Engine.

Euclid Score Euclid

Open Insurance Risk Navigator Score. A composite risk score (0–100) that combines Identity Risk, Fraud Signal, and Health Risk to assess overall underwriting risk. Higher scores indicate higher risk.

Identity Risk IR

Measures the legitimacy of the lead's identity based on phone validation, email deliverability, address verification, and name consistency. Higher scores suggest identity fraud or data quality issues.

Fraud Signal FS

Detects patterns indicative of fraud, including duplicate submissions, suspicious timing, velocity checks, and agent behavior anomalies. Higher scores indicate likely fraudulent activity.

Health Risk HR

Estimates the medical risk profile based on age, pre-existing conditions, medication history, and predictive health models. Used for underwriting assessment and rate determination.

Enhanced Provider Fraud Score EPFS

Advanced fraud detection score that analyzes provider network patterns, billing anomalies, prescription patterns, and cross-reference medical claim data to detect sophisticated healthcare fraud schemes. Scores range from 0–100.

Agent Pattern Risk APR

A composite score (0–100) that evaluates an agent's submission patterns for fraud indicators, including high-risk lead concentration, duplicate rates, VOIP usage, OTP failures, and submission burstiness.

Voice over IP VOIP

Internet-based phone numbers (e.g., Google Voice, Skype) that are not tied to a physical location. Often associated with higher fraud risk as they are easy to obtain and discard.

One-Time Password OTP

A temporary verification code sent via SMS or email to confirm identity. High OTP failure rates can indicate fake contact information or fraud attempts.

Submission Burstiness Burstiness

A statistical measure of temporal clustering in agent submissions. High burstiness (> 1.5) suggests unnatural submission patterns, potentially indicating bot activity or coordinated fraud.

Phone Validation PV

Real-time verification of phone number validity, type (mobile/landline/VOIP), carrier, and activity status. Critical for identity verification and contact quality.

Duplicate Lead Duplicate

A lead submission that shares key identifying information (name, DOB, SSN, address) with another lead in the system. May indicate data entry errors or deliberate fraud attempts.

Portfolio View Portfolio

Aggregate risk metrics across all leads in the current dataset. Provides executive-level insights into overall portfolio health, trends, and risk distribution.

Product Type Product

The insurance product category (ACA, Medicare, Medicaid, Life, Dental). Different products have varying risk profiles and regulatory requirements.

Affordable Care Act ACA

Health insurance marketplace plans mandated by the Affordable Care Act. Subject to specific enrollment periods, subsidies, and regulatory requirements.

Medicare Medicare

Federal health insurance program primarily for people age 65 and older, or younger people with qualifying disabilities. Includes Part A (hospital), Part B (medical), Part C (Advantage), and Part D (prescription drugs).

Medicaid Medicaid

State and federal program providing health coverage to eligible low-income individuals and families. Eligibility and benefits vary by state.

Underwriting UW

The process of evaluating a lead's risk profile to determine eligibility, coverage terms, and premium rates. Euclid provides automated risk signals to support this process.

High Risk Tier ≥ 70

Requires manual review, additional verification, or potential rejection due to significant identity, fraud, or health concerns.

Medium Risk Tier 40–69

Requires standard underwriting review with attention to flagged issues. May be approved with conditions or rate adjustments.

Low Risk Tier < 40

Passed all verification checks with no significant fraud or health risk signals. Generally fast-tracked for approval.