The problem nobody solves
Your inbox contains messages that demand action - fines, payment demands, fraud attempts. Most of us find them too late.
Critical messages buried in noise
A fine, a debt demand, or a fake notice - all drowning in the flood of email.
No way to tell what's real
A fraud message looks exactly like a real one. Without analysis tools, you're guessing.
The response is always late
By the time you find the problem, the window to act has usually closed.
How MAGEN works
Three steps. No complexity.
Connect
Secure connection to your Gmail via OAuth. Takes 30 seconds. No passwords.
MAGEN Analyzes
Every message is scanned across 6 layers: sender, content, links, attachments, timing, and thread context. 3 independent scores per message.
You Decide
The dashboard shows a clear state: green, orange, or red. Every alert is explained. The action is yours.
Brain 2.0 - Deterministic Analysis Engine
No guessing. No vague "artificial intelligence." Every decision is traceable.
MAGEN's analysis engine runs every message through 5 fixed layers. The result: 3 independent scores - Trust, Risk, and Fraud. Every score point is explained.
Signal Extraction
6 signal categories identified from every message
Scoring
3 independent scores: Trust, Risk, Fraud
Aggregation
Weighted by category and structural red flags
Decision
Status determined: clear, review, or action needed
State Management
Dashboard and alerts updated
- Deterministic rules - same input always returns the same result
- Structural red flags - instant override for seed phrase requests, executable attachments
- Bulk/marketing detection - suppresses false positives on mass emails
- Bilingual analysis - keyword detection in both Hebrew and English
An engine that monitors itself
MAGEN doesn't just analyze messages - it continuously measures its own performance.
Brain Health
7-day rolling window. Tracks detection rates, accuracy, and stability.
Score 0–100System Health
Response time, memory, uptime, and load monitoring.
Healthy / Monitor / CriticalServices Status
Simultaneous monitoring of 4 services: site, admin panel, app server, database.
4/4 activeThe mobile experience
Needs review at a glance
When something needs attention, the dashboard turns orange with a clear summary. One tap to see the details.

Full alert breakdown
Every alert shows fraud score, risk score, decision, and recommendation. No black boxes.

Complete alert history
All past alerts in one place. Search, filter by date, and review the full analysis anytime.

Suspicious text check
Got a strange SMS or message? Paste it directly and get an instant assessment.

Full transparency. No surprises.
MAGEN was built on one principle: you should know exactly what happens with your data.
Security
- Secure Gmail connection via OAuth
- No password storage - ever
- Revoke access anytime from Google settings
Privacy
- No data selling
- No invasive tracking
- Minimal data only - just what's needed to operate
Transparency
- Alert reasoning is fully visible
- Every score can be decomposed - you can always see why
- Analysis code is deterministic - doesn't change between runs
What MAGEN does for you
Deep Alert Analysis
Every suspicious message is analyzed with a clear risk score, detailed findings, and a tailored recommendation.
Suspicious Text Check
Paste any suspicious message and get an instant assessment - whether it's safe or needs caution.
Real-Time Fraud Detection
MAGEN detects suspicious messages and alerts you immediately - before you have a chance to respond.
What happens when MAGEN detects something
MAGEN analyzes each message, presents findings transparently, and lets you choose what to do.
Detection & Analysis
Each message is scanned for fraud indicators, urgency, and suspicious content. Results include a clear score and recommendation.
Actions in your hands
Moves the alert to history as resolved.
Prevents the alert from reappearing in the active list.
Alert History
All alerts are saved. You can search, filter by date, and view the full analysis at any time.
