AI

INSURANCE FRAUD
IDENTIFICATION

Fraud Detection You Can Trust

AI

INSURANCE FRAUD
IDENTIFICATION

Fraud Detection You Can Trust

DAFNE is an artificial intelligence platform that analyzes damage in accident images to detect large-scale fraud.

Estimated value of potential frauds identified by DAFNE to date
0 million €

How it works

Data transmitted
Claims
SX-2026-04871
Date
14/03/2026
Vehicle
Alfa Romeo Stelvio
01 / 06

Ingestion of the claims

Every day the company sends claims to be verified according to its operational criteria: value, geographic area, sampling, and alerts.

02 / 06

Automatic anonymization

The images are stripped of sensitive elements—license plates and faces. DAFNE is interested in the damage, not the identity.

License plate removed
0.97 · Broken Glass
Damage analyzed
Type
Broken Glass
Extension
0.31 m²
Signature
8F2A·4C19
03 / 06

Damage analysis

DAFNE identifies and isolates the damage, analyses its shape, extent and typology, and constructs a technical representation comparable over time.

04 / 06

Matching with the historical

The damage is compared with the company’s archive to reveal recurrences and reuses between claims even distant in time.

0.94 · Same damage
Left match data
Claim
SX-2025-03562
Date
12/01/2025
Vehicle
Alfa Romeo Stelvio
Claim under review
Dettaglio del lunotto infranto del sinistro in verifica
Historical record
Dettaglio del lunotto infranto del riscontro d’archivio

Match confirmed by M. Rossi

05 / 06

Human validation

High-confidence matches are verified by specialized analysts: false positives are eliminated and the case is confirmed. Human intervention is focused only where necessary.

06 / 06

Report generation

If confirmed, a structured evidence report is produced, with a direct comparison between the damage and traceable references.

Evidence report DAFNE con confronto tra il sinistro in verifica e il riscontro di archivio
M. Rossi Analista senior

Four levels of analysis

Every claims leaves traces. Not only in the declared dynamics, but also in the image pixels, file metadata, and attached documents. DAFNE reads them, correlates them, and compares them with historical records. The four levels do not operate in isolation: the evidence cross-references and reinforces each other. No result is sent to the insurance company until validated by an expert operator.

Pixel Analysis

From fraud to risk

Behind an image lies information the naked eye can’t see: when it was taken, the device used, and whether it has been altered. DAFNE detects discrepancies between the date of the shot and the date of the accident, traces of editing, images generated or retouched using artificial intelligence, and inconsistencies within the metadata chain.

Metadata Analysis

Each file tells how it was born.

Behind an image lies information the naked eye can’t see: when it was taken, the device used, and whether it has been altered. DAFNE detects discrepancies between the date of the shot and the date of the accident, traces of editing, images generated or retouched using artificial intelligence, and inconsistencies within the metadata chain.

Document Analysis

The paper also has cracks.

Estimates, invoices, forms, and appraisals can contain anomalies that are difficult to manually detect. DAFNE detects changed headers, fictitious names and contact details, recurring phone numbers in seemingly unrelated cases, and reused documents with minor variations from one claim to the next.

Satellite image analysis

The claim in its context.

Thanks to the partnership with LambdAI Space, the analysis can extend beyond the vehicle to include buildings, roofs, and outdoor areas. This is a particularly useful tool in property management, where photographic documentation is often partial and acquired only after the event.

The Visada Ecosystem

Clients
Partner
Investors and Financing

Awards and recognitions

Ecosystem Impact Awards 2026
Best Insurtech 2026
Winner “7 Ideas to change Italy”
Best Innovative Startup Southern Italy 2025

Explore DAFNE

Book a demo to see DAFNE in action

Find out how DAFNE analyzes claims, identifies anomalies, and provides validated evidence that can be directly used by anti-fraud teams. Enter your email address: our team will contact you to arrange a dedicated demonstration.