FI
Platform / Insurance Fraud Intelligence
Production

Home

Command center for this workspace — live status across every module and model.

Q2 2026 · All lines
184K
Claims screened (30d)
3.8%
Suspicious claim rate
₹214 Cr
Leakage prevented (YTD)
6
AI models in production
Fraud intelligence modules
Six connected modules, one fraud intelligence graph. Click any tile to open it.

AI models powering this workspace

Claim Fraud ScorerAUC 0.88
Fraud Ring DetectorPrecision 0.83
Provider Collusion ModelAUC 0.85
Document Forgery DetectorShadow testing
AI Insights
Generated
₹38 Cr of expected leakage sits in 2.1% of claims linked to 14 suspected fraud rings — prioritize these for SIU.
Motor claims filed within 30 days of policy inception are 4.6× more likely to be fraudulent.
Three repair garages account for 22% of flagged motor claims this quarter — the Provider Collusion model recommends network review.
184K
Claims screened (30d)
3.8%
Suspicious claim rate
₹214 Cr
Leakage prevented (YTD)
1,860
Open SIU cases

Leakage prevented (₹ Cr), trailing 6 months

0 20 40 60 Jan Feb Mar Apr May Jun

Straight-through processing — live gauge

0%
of claims auto-cleared by AI
rest routed to review or SIU

Flagged claims by fraud type

Inflated claim38%
Staged / false incident24%
Provider collusion18%
Identity / document12%
Early claim8%

Suspicious claim rate, by line of business

Motor
5.1%
Health
4.3%
Property
2.8%
Travel
2.1%
Marine
1.4%
Claim Fraud Review — CLM-40871
Motor · Own Damage  ·  Incident 12-06-2026  ·  Reported 14-06-2026  ·  Policy age: 22 days
Fraud score: 82 / 100 · High risk
Policy age at claim — filed 22 days after inception
22 days
Watch
Repair garage — on SIU watchlist, 3 prior flagged claims
Watchlisted
Flag
Document consistency — repair estimate vs damage photos
Mismatch
Flag
Prior claims — claimant, last 24 months
4 claims
Watch
Claim amount vs sum insured — near policy limit
94%
Watch
Police report — FIR filed for the incident
Filed
Clear
AI Fraud Copilot
Claim Fraud Scorer · v5.2
Live
Recommended: Refer to Special Investigations Unit (SIU) — do not fast-track 86% confidence

The model estimates an 82% fraud probability and links this claim to ring RING-0091 through repair garage G-88. Expected leakage if paid as-is is ₹3.8L; comparable referred claims resolve with around ₹2.9L saved through denial or negotiated settlement.

Why the model flags this claim — top signal contributions
Repair garage on watchlist
+0.34
Estimate / photo mismatch
+0.27
Early claim (22 days)
+0.19
4 prior claims in 24 months
+0.14
Police FIR filed
−0.12
42
Active fraud rings
1,240
Claims linked to rings
₹68 Cr
Exposure inside rings

Fraud ring RING-0091 — link analysis

Five claims connected through shared entities — one repair garage, one hospital, a phone number, a bank account, and an agent. The AI flags the cluster as a single organized ring.
G-88 Phone Bank H-21 A-330 40871 40912 41003 41190 41255
Anchor provider (repair garage) Shared entity Linked claim

Most common shared links across rings

Same repair garage
100%
Same phone number
78%
Same bank account
61%
Same doctor / hospital
54%
Same agent
39%

Claims linked to ring RING-0091

ClaimLineShared linksAmountRole
CLM-40871MotorGarage G-88, Phone₹3,80,000Anchor
CLM-40912MotorGarage G-88, Bank₹2,90,000Linked
CLM-41003HealthHospital H-21, Phone₹1,60,000Linked
CLM-41190MotorGarage G-88, Agent₹4,10,000Linked
CLM-41255HealthHospital H-21, Bank₹2,20,000Linked
No rows match your search.
184K
Claims scored (30d)
82%
Auto-cleared
3.8%
Routed to SIU

Fraud-score simulator

Toggle the red-flag indicators present on a claim. The AI recomputes the fraud score and the routing decision in real time.
82/ 100 fraud score
Refer to SIU

How the score routes

Every claim is routed by its fraud score. The marker shows where the current simulation lands across the three decision bands.
03570100
Auto-approveManual reviewRefer to SIU

At 82, this claim exceeds the SIU threshold of 70 and is auto-referred for investigation before any payout.

Claim routing distribution (last 30 days)

Auto-approve & pay
82.0%
Manual review
14.2%
Refer to SIU
3.8%
24K
Providers monitored
1,240
Flagged for review
42
Collusion rings

Flagged providers by type

Repair garages
520
Hospitals / clinics
384
Diagnostic centres
270
Agents / intermediaries
198
Towing / salvage
108

Providers flagged for collusion review

ProviderTypeLinked claimsBilling vs peerAI signal
G-88Repair garage34+68%Ring link
A-330Agent22n/aRing link
H-21Hospital28+54%Inflated billing
DC-119Diagnostic centre19+47%Duplicate reports
No rows match your search.
1,860
Open SIU cases
940
Cases closed (30d)
₹18.6 Cr
Leakage prevented (30d)

SIU investigation & outcome audit trail

TimestampClaimActionBasisOutcome
07-05 09:14CLM-40871Referred to SIUFraud score 82, ring link RING-0091Under investigation
07-05 09:48CLM-39220Claim deniedStaged incident confirmed by field visit₹3.2L saved
07-05 10:20CLM-40012Cleared & paidInvestigation found no fraud indicatorsFair-handled
07-05 10:55CLM-41190Field investigation openedGarage G-88 ring link, photo mismatchUnder investigation
No rows match your search.
5
Models in production
1
In shadow testing
140ms
Avg. inference latency
920K
Predictions served today

Registered models

ModelTypeStatusPerformanceLast trainedExplainability
Claim Fraud ScorerGradient-boosted treesProductionAUC 0.883 days agoSHAP attribution
Fraud Ring DetectorGraph neural networkProductionPrecision 0.834 days agoSubgraph attribution
Provider Collusion ModelAnomaly + peer benchmarkingProductionAUC 0.856 days agoSHAP attribution
Claim-Text NLP ClassifierTransformer (claim narratives)ProductionF1 0.815 days agoToken attribution
Duplicate-Claim MatcherEntity resolution modelProductionPrecision 0.909 days agoMatch-feature scores
Document Forgery DetectorVision model (image forensics)Shadow testingAUC 0.791 day agoSaliency maps
No rows match your search.

Global explainability — Claim Fraud Scorer

Mean absolute SHAP contribution across all scored claims — what drives fraud scores overall.
Provider watchlist status
0.29
Estimate / photo consistency
0.25
Policy age at claim
0.20
Prior claim frequency
0.16
Claim-to-sum-insured ratio
0.12

Model lifecycle

01
Train
02
Validate
03
Shadow test
04
Promote
05
Monitor
Shadow models score live claim traffic in parallel without serving decisions, until they clear the production accuracy bar.
6
Connected systems
1.4M
Records synced today
0
Sync errors (24h)

Connected systems

SystemTypeStatusLast syncRecords today
Claims Management SystemFNOL, claim status & payoutsConnected3 min ago640K
Policy Administration SystemPolicy, inception & sum insuredConnected6 min ago320K
Provider / Network RegistryGarages, hospitals & agentsConnected9 min ago88K
Industry Fraud Registry (IIB)Cross-insurer fraud & claims dataConnected30 min ago140K
Document & Image StoreEstimates, photos & FIRsConnected4 min ago210K
Telematics / Device FeedMotor incident & GPS dataSyncingIn progress36K
No rows match your search.

Workspace

Workspace nameAegis General Insurance
EnvironmentProduction
Data residencyMumbai, IN · On-prem
API keysk-live-••••••••4f2a  

Team access

Rajiv Bhatia · Head of Claims FraudActive now
Meera Sundaram · SIU Lead40m ago
Fraud Scoring Bot · Service accountActive now