Isolation Forest ML Any Linux Distro Auto-Block Threats Multi-Tenant

AI Firewall Monitoring Platform

Machine learning anomaly detection trained on your own traffic. Detects threats signature-based tools miss. Pushes block rules directly to your firewall — automatically, in real time.

Get Platform Access Platform Login
Live Threat Detection
Port Scan Detected
185.220.101.x → 22, 80, 443
HIGH
Anomaly: Outbound Spike
Server-03 → unknown destination
MED
Auto-Blocked: Shodan Scanner
Rule pushed to OPNsense

What is AI Firewall Monitoring?

AI firewall monitoring is a network security methodology that replaces static signature databases with machine learning models trained on your network's own traffic patterns. Rather than asking "does this packet match a known-bad signature?", AI monitoring asks "is this connection statistically abnormal for this network?"

Traditional intrusion detection systems (IDS) rely on signatures — lists of known bad IPs, known exploit payloads, known malware hashes. They're effective against known threats, but blind to zero-day exploits, low-and-slow reconnaissance, insider threats, and novel attack patterns. An attacker who is slightly patient or slightly creative evades them entirely.

The GoTech AI Firewall Monitoring Platform applies Isolation Forest — a proven, production-grade unsupervised ML algorithm — to firewall logs, syslog, server auth logs, DNS query logs, and NetFlow/IPFIX data. The model is trained per-client, so it learns what "normal" looks like for your specific environment: your server talking to its usual databases, your employees connecting from their usual subnets, your firewall seeing its usual traffic mix.

When traffic deviates from that learned baseline — a new outbound destination that's never appeared before, a login from an unexpected country, a server process suddenly making DNS queries it's never made — the anomaly scores spike and an alert fires. High-confidence threats can trigger Auto-Apply mode, which pushes block rules directly to your firewall (OPNsense, pfSense, iptables, nftables) in real time.

How Isolation Forest Works
1
Train on normal traffic

The model ingests weeks of your baseline log data — firewall connections, server auth events, DNS queries, bandwidth patterns.

2
Random isolation partitioning

Isolation Forest randomly splits the feature space. Normal data points require many splits to isolate; anomalies are isolated quickly with fewer splits.

3
Score & act

Each new connection or event gets an anomaly score. Low scores = normal. High scores = flag for review or auto-block, depending on your settings.

Why unsupervised ML?

No labeled attack examples are needed. The model learns what "normal" is and flags deviations — meaning it catches brand-new attack types that no signature list has seen yet.

What Threats Does It Detect?

The platform combines rule-based detection for known threat categories with ML anomaly detection for unknown threats. Both run in parallel.

Port Scans & Reconnaissance

Detects sequential port sweeps, service banner grabbing, and slow-scan techniques designed to evade rate-based detection rules.

Brute Force & Credential Stuffing

SSH, RDP, VPN, and web application login brute force — including distributed attacks spread across multiple source IPs.

DDoS & Volumetric Attacks

SYN floods, UDP amplification, and ICMP flood detection with automatic rate-limit and geo-block rule generation.

Malware C2 Communication

Outbound connections to command-and-control infrastructure — detected via ML anomaly scoring on destination IPs/ports and DNS query patterns.

Data Exfiltration

Unusual outbound bandwidth spikes, large file transfers to unknown destinations, or high-volume DNS responses that indicate DNS-based data exfiltration.

DNS Tunnelling

Detection of DNS queries used to exfiltrate data or establish covert C2 channels — catches iodine, dnscat2, and similar tools.

Known Scanner IP Ranges

Automatic blocking of Shodan, Censys, Masscan, and other internet scanner infrastructure — stops reconnaissance before it generates noise.

Geo-Blocking

Block inbound traffic from specific countries or regions. Rule is pushed directly to your firewall. Easily adjustable per-device or globally.

ML Behavioral Anomalies

Any traffic that deviates significantly from your network's learned baseline — catches novel attack patterns before any signature list has ever seen them.

Supported Platforms

Full API integration means the platform can push block rules directly to your firewall — not just read logs.

Firewall Platforms

Firewall Integration Status
OPNsenseFull API + syslogFull Support
pfSenseFull API + syslogFull Support
Linux (iptables)fw-agent + APIFull Support
Linux (nftables)fw-agent + APIFull Support
Linux (ufw)fw-agent + APIFull Support
Cisco ASASyslog ingestionPartial (read-only)
Fortinet FortiGateSyslog ingestionPartial (read-only)
Palo Alto NetworksSyslog ingestionPartial (read-only)
SonicWallSyslog ingestionPartial (read-only)
Cisco FirepowerPlannedPlanned
Sophos XGPlannedPlanned
Ubiquiti UniFiPlannedPlanned

Linux Server & Desktop Agent (fw-agent)

Install command
curl -sSL https://ai077.gotech.us/install | sudo bash

Communicates outbound over HTTPS port 443 only — no inbound firewall rules needed on your server.

Distribution Status
Ubuntu (18.04+)Supported
Debian (10+)Supported
RHEL / CentOS (8+)Supported
Rocky LinuxSupported
Arch LinuxSupported
Alpine LinuxSupported
Amazon Linux (2/2023)Supported
openSUSE Leap / TumbleweedSupported
Generic Linux (any systemd)Supported

How It Works

From zero to AI-monitored in under 30 minutes.

1
Sign up & create your tenant

Submit the quote form — we provision your isolated tenant on the platform within one business day. You get a login at ai077.gotech.us and access to your dashboard.

2
Connect your firewall(s)

For OPNsense or pfSense: configure the syslog UDP destination to point to the platform collector, or provide API credentials. For Linux firewalls: point your existing syslog facility at us.

3
Install the fw-agent on servers

Run a single curl | sudo bash command on each Linux server or workstation you want monitored. The agent ships logs outbound over HTTPS/443 — no inbound holes needed.

4
The ML model trains on your baseline (24–48 hours)

The Isolation Forest model ingests your actual traffic. After 24–48 hours it begins scoring connections against your specific normal patterns. Rule-based detection (port scans, brute force, known scanners) fires immediately from day one.

5
Review alerts — or enable Auto-Apply

Threats appear in your dashboard with severity scores, source IPs, and context. Approve manually to push block rules, or enable Auto-Apply for high-confidence threats. Every action is logged with full audit trail.

Simple Per-Device Pricing

No long-term contracts. No setup fees. Add or remove devices any time. Discounts available for home users and education/non-profit organizations.

Firewall
$35/mo

Per firewall device. Full API integration — rule push, log ingestion, auto-block.

Most Common
Server
$20/mo

Per Linux server. fw-agent installed, auth log + syslog + iptables monitoring.

Desktop / Workstation
$10/mo

Per endpoint. Lightweight fw-agent monitoring for Linux workstations.

Miscellaneous
$30/mo

Per device. IoT, NAS, specialty appliances — via syslog or custom log shipping.

Get a Quote

Home user and education/non-profit discounts available. Contact [email protected]

Frequently Asked Questions

AI firewall monitoring uses machine learning to analyze real-time firewall and server logs, detecting anomalous network behavior that signature-based tools miss. The GoTech platform uses Isolation Forest unsupervised ML trained on each client's own traffic patterns, enabling it to flag deviations like unusual outbound connections, unexpected login behavior, or low-and-slow port scans — without relying on known-bad signature lists.

Isolation Forest is an unsupervised ML algorithm that detects anomalies by randomly partitioning data points. Abnormal data points — like unusual IP connections or unexpected traffic spikes — are isolated quickly with fewer random splits, giving them a lower anomaly score. For network security, the model trains on normal traffic patterns and flags connections that deviate significantly. Because it's trained per-client, it minimizes false positives compared to shared threat feeds.

Full API support (including rule push and auto-block) for OPNsense, pfSense, and Linux firewalls (iptables, nftables, ufw). Partial read-only support for Cisco ASA, Fortinet FortiGate, Palo Alto Networks, and SonicWall. Cisco Firepower, Sophos XG, and Ubiquiti UniFi are planned.

The fw-agent supports Ubuntu, Debian, RHEL, Rocky Linux, Arch Linux, Alpine Linux, Amazon Linux (2 and 2023), openSUSE, and Generic Linux (any systemd-based distro). It installs with a single curl | sudo bash command and communicates outbound over HTTPS port 443 only — no inbound ports or firewall exceptions needed on your server.

Per-device monthly pricing: $35/month per firewall, $20/month per server, $10/month per desktop/workstation, and $30/month per miscellaneous device. No contracts, no setup fees, no minimum device count. Cancel or add devices any time. Home user and education/non-profit discounts available.