NonaShield

Diime AI NonaShield Platform is a mobile fraud detection and device trust platform that operates across nine architectural layers, combining on-device runtime protection with backend intelligence services delivered through a single SDK and management console.

On-device layers (five signal layers):

• Phone Health: Detects root, tamper, hooking, overlays, and debuggers via Runtime Application Self-Protection (RASP)
• Device Recognition: Persistent device identification across sessions and accounts
• Hardware Identity: Uses the device security chip to cryptographically sign transactions
• API and Request Integrity: Certificate pinning, proxy detection, and end-to-end signed payloads
• Behavioural Analysis: Session behaviour scoring, contributing 20-25% of the overall risk score

Backend layers (four intelligence layers):

• Risk Score and Fraud Engine: Scores only signature-verified inputs, producing a unified score across app, web, and partner channels
• Graph Analysis: Links device, account, beneficiary, and session data to surface fraud rings and shared-device patterns
• Agentic AI Advisory: Generates plain-language explanations of decisions, policy recommendations, and pre-drafted analyst case notes
• Cryptographic Evidence: Produces a signed, tamper-evident, exportable record per transaction for dispute resolution

The platform covers 400+ threat vectors, 100+ RASP detection vectors, and 47+ behavioural vectors. All signals leaving the device are cryptographically signed before being scored. The platform explicitly acknowledges its limitations: it cannot prevent authorised push payment fraud where a legitimate user on an enrolled device initiates a transfer, but it records device and app integrity status, location, and device identity at the time of signing for use in subsequent dispute and investigation processes.