Scam Intelligence
Instantly analyze suspicious WhatsApp forwards, SMS, or Telegram messages for fraud signals.
Try an example:
0 / 5000
Processed Locally • No Data Stored
Detection Pipeline Architecture
01
Rule Engine
100+ Regex patterns fire instantly (<50ms). Scans for urgency, financial, and authority impersonation signals.
02
Score Threshold
Risk Score < 15: Safe. Score 15–39: Caution. Score ≥ 40: Escalate to secondary AI analysis.
03
LLM Heuristics
For high-score cases, Claude Haiku provides advanced contextual reasoning to eliminate false positives.