Six-layer forensic analysis of every loan document — PDF metadata, payroll maths, live ABN checks, and cross-reference verification — in under 60 seconds.Mortgage, personal, and business lending.
In February 2026, Commonwealth Bank self-reported ~A$1 billion in suspected fraudulent mortgage applications — fake payslips and bank statements generated with AI tools, submitted through broker channels. Westpac and ANZ have since flagged similar issues.
Analyses creator metadata, font fingerprints, modification timestamps, and embedded image manipulation — hallmarks of AI-fabricated documents.
Claude Sonnet semantically reads each document for AI-generation patterns, terminology anomalies, and field inconsistencies specific to Australian payroll.
Live ABN Lookup, ASIC company register, BSB directory, and ABS wage benchmark checks. Invalid ABNs and non-existent employers caught instantly.
Deterministic checks: gross − tax = net, super at 11.5% SGC rate, YTD consistent with pay periods elapsed since 1 July, impossible dates caught.
Tracks submission velocity, fraud rates, and shared-employer network clustering across all applications submitted by each broker.
Every risk score comes with a plain-English narrative and specific evidence — meeting ASIC and APRA explainability requirements for loan decisions.
Start free. Scale when you're ready. Also available: $15/case pay-as-you-go API.