SilverGuard

Overview

SilverGuard is a comprehensive security system designed to protect senior citizens from financial fraud and digital scams. It monitors communication channels in real-time to detect suspicious activities and alerts caregivers immediately.

The system combines a Flutter mobile app with an n8n automation backend to create a multi-layered protection pipeline. Incoming calls and messages are screened in real-time against known spam databases. When a potential threat is detected, the system instantly notifies family members and caregivers through multiple channels, providing context about the threat so they can take appropriate action.

Key Features

• Real-time call and SMS monitoring for fraud detection patterns
• Automated multi-channel alerts to caregivers via Twilio API (SMS + voice)
• Spam number identification and scoring using RapidAPI phone lookup
• User-friendly Flutter dashboard for activity monitoring and threat history
• Secure family connection system for managing trusted contacts
• n8n workflow automation for orchestrating detection and notification pipelines

Architecture & System Design

SilverGuard uses a four-layer architecture optimized for real-time responsiveness:

• Mobile Layer: A Flutter/Dart cross-platform app intercepts incoming calls and SMS events, extracts caller metadata, and pushes events to the backend via Firebase Cloud Messaging for instant delivery
• Automation Layer: n8n workflows orchestrate the threat detection pipeline — receiving events, routing them through API checks, applying scoring rules, and triggering notification sequences without custom server code
• API Layer: RapidAPI phone number lookup identifies known spam callers, while Twilio handles outbound alert delivery (SMS + automated voice calls) to registered caregiver contacts
• Data Layer: Firebase Firestore stores user profiles, trusted contacts, threat logs, and historical activity data with real-time sync across devices

Challenges & Solutions

• API Rate Limits: Phone number lookup APIs have strict rate limits. Implemented a local cache of previously checked numbers with TTL-based expiration, reducing API calls by 70% while maintaining fresh data
• False Positive Alerts: Overly aggressive spam detection triggered alerts for legitimate calls. Added a whitelist system with family contacts and frequency-based scoring that gradually learns safe numbers
• Real-time Latency: Initial alert delivery took 8-10 seconds. Switched from polling to Firebase Cloud Messaging push delivery, reducing alert latency to under 2 seconds

Screenshots