Gateway Sentry
Looking GlassDashboard

Products / Web Protection / AI Threat Detection

Detection That Learns Your Traffic

Machine learning models your application's normal behavior, then scores every request at the edge using behavioral, fingerprint, geographic, and velocity signals.

seconds to detect deviationpassive mode by default0 extra pricing tiers
Get ProtectedWeb Protection overview

How the AI Responds

Baseline Modeling

Models continuously ingest the live event stream to learn your traffic: baselines, fingerprint distributions, and threat patterns. Normal is defined by your application, not by a generic template.

Request Scoring

Every request is scored at the edge with behavioral, fingerprint, geographic, and velocity signals fused into a composite threat score. No single signal decides; the correlation does.

Graduated Response

Passive by default: traffic flows while the AI observes. Responses escalate from monitoring to challenges to blocking, keeping false positives near zero.

Zero-Downtime Updates

Updated detection models propagate to all edge nodes with zero-downtime refreshes, so protection always runs the latest intelligence without a maintenance window.

Passive First, Always

The AI earns enforcement rights by observing: traffic flows untouched while it learns your baseline, and responses only escalate as confidence does. Protection that starts by blocking is protection you end up turning off.

The Network Effect

Every new pattern that appears anywhere on our network becomes part of how we protect everyone else on it. Your quiet Tuesday benefits from someone else's bad one. More on how we think about this on the About page.

Keep Exploring

Bot Protection

Where AI scoring meets fingerprints and challenges.

WAF

Turn model verdicts into route-level policy.

Web Protection

The application-layer product AI detection ships inside.

Network Protection

The L3/L4 layer underneath, judging packets instead of requests.

Frequently Asked Questions

What is AI threat detection?
AI threat detection uses machine learning to model your application's normal traffic, then scores every request at the edge using behavioral, fingerprint, geographic, and velocity signals. Gateway Sentry detects deviations from baseline within seconds and neutralizes threats before they reach your infrastructure.
How does Gateway Sentry's AI detect attacks without blocking real users?
The system runs in passive mode by default, so all traffic flows to your origin while the AI silently observes and scores it. Only requests whose fingerprints fail trust evaluation are routed to a challenge layer, and responses escalate gradually from monitoring to challenges to blocking, which keeps false positives near zero.
Can AI threat detection stop phishing and bot traffic?
Yes. Gateway Sentry fuses fingerprint intelligence, protocol behavior, and spoofing detection to catch clients pretending to be something they are not, including phishing infrastructure, scrapers, and credential-stuffing bots. Multi-signal correlation produces a composite threat score that triggers an automatic response.
How quickly does the AI adapt to new attack vectors?
Models continuously ingest the live event stream and update traffic baselines, fingerprint distributions, and threat patterns. Updated detection models propagate to all edge nodes with zero-downtime refreshes, so your protection always runs the latest intelligence.
What does AI threat detection cost?
AI detection is built into Gateway Sentry's usage-based protection. L7 web protection is $30 per TB plus $0.55 per million requests, and L4 network protection is $3.50 per TB plus $4.39 per IP, billed pay-as-you-go with no minimums.

Get Protected in Minutes

Point your DNS at the edge and deploy rules from one dashboard. No code changes.

Get Protected