AI-driven Network Detection & Response
See the attacks rules don’t catch
AXIUM IDS is a passive network sensor that runs machine-learning detectors on your traffic. It uncovers C2 beaconing, data exfiltration, lateral movement and other attacker behaviour — even in encrypted traffic, without decrypting it.
- 100% on-premise
- No agent
- No decryption
10.20.4.17 · finance-ws-07- T1568.002 Domain generation algorithm detected in DNS queries
- T1071 Periodic C2 beaconing over TLS to a rare destination
- T1021 New remote-access sessions to 3 internal servers
- T1048 Data exfiltration pattern over DNS
Illustrative example
Why AXIUM
Signatures see what they already know.
Attackers don’t play by the rulebook.
Most traffic is encrypted
Payload inspection goes blind on TLS. AXIUM learns from metadata, TLS fingerprints and packet-size patterns instead.
Attackers change infrastructure
New domains, new IPs, new tools. Behavioural models catch the technique — beaconing, scanning, exfiltration — not the indicator.
Analysts drown in alerts
AXIUM correlates isolated alerts into campaigns per host, so your SOC investigates incidents, not noise.
Observe
Passive copy of your traffic from a TAP or SPAN port.
Detect
ML detectors score each flow, host and session.
Correlate
Related alerts are fused into one campaign per host.
Respond
Alerts with ATT&CK context go straight to your SIEM / SOAR.
Detection coverage
ML detectors mapped to MITRE ATT&CK
Each detector targets a specific attacker technique. Every alert tells your analysts what is happening and where it sits in the kill chain.
C2 beaconing
Periodic call-backs to command & control, including over TLS.
DGA domains
Algorithmically generated domains used by malware.
Exfiltration
Data smuggled out through DNS or ICMP tunnels.
Malicious TLS
Suspicious encrypted channels, spotted from handshake and certificate features.
Lateral movement
Unusual remote-access sessions and DCERPC-based pivoting.
Kerberoasting
Service-ticket harvesting against Active Directory.
Brute force
Password guessing against exposed services.
Slow scans
Low-and-slow reconnaissance that stays under thresholds.
DoS / DDoS
Network and application-layer denial of service.
Webshells
Web-server backdoors in cleartext HTTP traffic.
OT / SCADA
Anomalous Modbus and DNP3 commands to PLCs and RTUs.
YARA & threat intel
Malware signatures and known-bad TLS fingerprints.
How it works
Plug in a mirror port. Get alerts in your SIEM.
→
→
1. Mirror
Connect the sensor to a TAP or SPAN port. It is strictly passive: it never injects or blocks traffic, so it is safe even on industrial networks.
2. Detect
The sensor parses dozens of protocols (HTTP, TLS, DNS, SMB, Kerberos, SSH, Modbus, DNP3…) and runs ML models locally. Per-host baselines adapt to your environment.
3. Respond
Alerts and correlated campaigns are exported with their ATT&CK technique, host and evidence — ready for triage in the tools your team already uses.
Encrypted traffic
Visibility without breaking encryption
AXIUM does not need your private keys or a TLS-terminating proxy. It analyses what is still visible on the wire.
- TLS client and server fingerprints (JA3 / JA4) checked against threat intelligence
- Packet-size and timing patterns that reveal beaconing and tunnels
- Certificate, SNI and flow metadata fed to dedicated ML models
- Privacy-friendly: payloads stay encrypted, data never leaves your site
Who it’s for
One engine, several ways to use it
Security teams & SOC
Add a network detection layer next to your EDR and SIEM, with alerts your analysts can act on.
MSSP
Deploy a sensor per customer and feed your SOC platform with ATT&CK-labelled detections.
Industry & OT
Passive-only monitoring of Modbus and DNP3: nothing is ever injected into the process network.
Security vendors (OEM)
Embed the detection engine in your firewall, probe or SASE platform through a C library.
Deployment
Runs where your data is
All detection runs locally. No cloud dependency, no traffic sent outside your perimeter.
On-premise sensor
Install on your own hardware, connected to a TAP or SPAN port.
Virtual appliance
Ships as a container or VM image for your hypervisor.
Air-gapped
Offline signed licence: no call-home, suited to isolated and sensitive networks.
Offline PCAP analysis
Point it at a capture file for threat hunting or post-incident forensics.
Integrations
Fits into your existing SOC
No new console to learn. AXIUM sends structured alerts to the tools your team already watches.
Early access
Become a design partner
We are working with a small number of organisations to validate AXIUM on real networks. Test it on your own traffic and shape the product with us.
- Pilot on your network — or start with a PCAP you send us
- Direct line to the engineers who build the detectors
- Influence on the roadmap and founding-partner terms
FAQ
Frequently asked questions
Does AXIUM decrypt TLS traffic?
No. It analyses metadata that stays visible on encrypted connections: TLS fingerprints, certificates, SNI, packet sizes and timing. You do not need to share keys or deploy a decryption proxy.
Does any data leave my network?
No. Capture, analysis and ML inference all run on the sensor. Only the alerts you choose to export go to your own SIEM.
Can it block attacks?
AXIUM is a passive detection sensor: it never injects or drops traffic. Response is driven by your SOC tools (SOAR, firewall, EDR) using AXIUM alerts.
How does it differ from a signature IDS?
Signature-based IDSs only match known indicators. AXIUM models attacker behaviour — beaconing, tunnelling, lateral movement — so it can flag new infrastructure and tools. It also runs YARA and threat-intel lookups for known threats.
Is it suitable for industrial networks?
Yes. The sensor is strictly passive and includes dedicated models for Modbus and DNP3 traffic.
How can I try it?
The quickest way is to send us a PCAP capture: we run it through AXIUM and walk you through the results. You can also run a pilot on a mirror port of your network.
Let’s talk about your network
Questions, demo request or pilot: write to us and we will get back to you quickly.