SecureEdge AI — Email Security, DLP & Web Filtering | Madarson IT
Live on Azure · AWS · Google Cloud

It reads
the email.
Then explains.

SecureEdge AI scans every email and web request through Anthropic Claude, scores it 0–100, and blocks what's dangerous — telling you exactly why, in plain English. Self-hosted. No per-scan fees.

3 Clouds
0–100 AI threat score
50K+ IDS rules
$0 Per-scan fees

Most filters give you a score.
This one gives you a reason.

Here is a Business Email Compromise attempt arriving at a protected domain — and what SecureEdge AI did with it, in the milliseconds before it reached an inbox.

INBOUND · SCANNED BY SECUREEDGE-AI MILTER · CLAUDE ANALYSIS
From "Michael Chen, CEO" <ceo@evil-phishing.com>
To accounts.payable@yourcompany.com
Subject Urgent Wire Transfer — Confidential
Hi — I'm in back-to-back meetings and can't take calls.
Please wire $50,000 to account 4532015112830366 before close of business today.
Keep this confidential until the announcement. I'm counting on you.

— Michael
Threat score
95/100
BEC Impersonation Wire fraud
Why it was blocked: The sender claims to be an executive but the domain doesn't match your organisation. The message combines urgency, a request for secrecy, and a wire transfer to an unrecognised account — the signature pattern of Business Email Compromise.
✓ Blocked at SMTP ✓ Quarantined ✓ Sender notified ✓ Logged to dashboard

Everything inspected.
Nothing phoning home.

SecureEdge AI runs entirely inside your own cloud account. Your mail and traffic never leave your tenancy — the only outbound call is to the Anthropic API, using your own key.

Port 25 · Postfix milter

Email Security & DLP

Every inbound and outbound message is analysed by Claude for phishing, BEC, malware, spam and social engineering — then blocked, quarantined or allowed. Outbound mail is inspected for PII, PHI, PCI data and credentials before it leaves.

Both editions
Port 3128 · Squid proxy

Web Filtering & DLP

All HTTP/HTTPS traffic is scanned for malicious URLs, command-and-control domains, and data exfiltration through web uploads. Blocked users see a branded block page instead of a dead connection.

Complete only
Inline · Suricata IDS

Intrusion Detection

50,000+ Emerging Threats rules plus custom SecureEdge detections, updated automatically every week at no extra cost. Alerts surface on the same dashboard as your email and DLP events.

Complete only

Security you can
actually operate.

Built for teams who need enterprise-grade protection without a dedicated SOC, a rules language, or a per-message invoice.

Plain-English policies

Describe what you want — "block outbound email containing credit card numbers" — and Claude writes the working policy. No rule syntax to learn.

AI Security Assistant

Ask questions the way you'd ask a colleague. Summarise today's threats, build a policy, or explain a compliance gap.

Compliance-aware

Violations are mapped to the framework they touch — HIPAA, PCI-DSS, GDPR, SOC 2 and NIST — not just flagged as "sensitive".

Unified dashboard

Email, web and DLP events in one view, with quarantine management and a searchable threat history.

Role-based access

Admin, operator and auditor roles, so analysts can work and auditors can review without touching configuration.

Bring your own key

Claude runs on your own Anthropic API key, billed directly to you. No markup, no per-scan pricing, no vendor in the middle.

Works with your mail

Point your MX record at the appliance. Microsoft 365, Google Workspace, Exchange or on-premises — no migration required.

Cloud-native logging

Optional forwarding to Microsoft Sentinel, Amazon CloudWatch or Google Cloud Logging, authenticated by managed identity or instance role.

Hardened base

Madarson IT Hardened Ubuntu 24.04, TLS 1.2+ enforced, unique per-instance credentials generated on first boot. No default passwords.

Start with email.
Add the rest when ready.

Both editions are the same appliance and the same dashboard — Complete simply adds the web and intrusion-detection engines.

Edition 01

Email Security & DLP

The AI email gateway and data loss prevention platform. For teams whose biggest exposure arrives in the inbox.

  • AI email gateway — phishing, BEC, malware, spam
  • Outbound DLP — PII, PHI, PCI, credentials, IP
  • Plain-English policy builder
  • AI Security Assistant
  • Dashboard, quarantine, RBAC
Edition 02 · Flagship

SecureEdge AI Complete

Email, DLP, web filtering and intrusion detection unified in a single VM — replacing several appliances with one.

  • Everything in Email Security & DLP
  • Squid web proxy with AI DLP scanning
  • Suricata IDS — 50,000+ threat rules
  • Branded block page for filtered traffic
  • Unified email + web + DLP dashboard

Built for the teams
holding the line.

Healthcare

Stop PHI leaving through email or web uploads, with every violation mapped to the HIPAA safeguard it touches.

Financial services

Block wire-fraud and CEO-impersonation attempts before they reach accounts payable. PCI-DSS, SOX and GLBA aware.

Legal

Protect privileged documents and client confidentiality across every outbound channel, not just the ones you remember to check.

SMBs without a SOC

Enterprise-grade email security that explains itself — so the person running IT can act on a threat without a security degree.

MSSPs

Deploy per client in their own tenancy. No per-scan fees, no data leaving their cloud, private offers available.

Consolidating teams

Replace a mail gateway, a web proxy and an IDS with one appliance and one dashboard — and one thing to patch.

Deploy on your cloud.

Same appliance, same dashboard, same Claude analysis — running inside your own account on all three major clouds. Launch, add your API key, point your MX record.

Responsible use

SecureEdge AI is for authorised deployment only. Operators are responsible for complying with applicable data-protection, email-interception and web-monitoring laws in their jurisdiction. For production mail, assign a static IP and configure reverse DNS (PTR), SPF, DKIM and DMARC.

Notices

Anthropic Claude is a third-party service requiring your own API key, billed to your Anthropic account and subject to Anthropic's terms. Ubuntu is a registered trademark of Canonical Ltd. Included open-source components (Postfix, Squid, Suricata and others) are provided under their respective licences; Madarson IT does not relicense them. For private offers, volume licensing or MSSP arrangements, contact info@madarsonit.com.