Platform

One hunting platform, five layers

Lairhunt joins detection, investigation, response, edge sensing and simulation into one loop. Every layer shares the same telemetry, entity graph and threat-behaviour corpus.

Architecture

What runs where

LayerWhat it doesRuns onProduct
Telemetry ingestTaps, agents and cloud connectors stream network, identity, endpoint and cloud data, with optional DPU capture.BlueField DPU, collectorsHuntra, Sentix
Detection pipelinePer-tenant behavioural baselines and ATT&CK classification over raw events.H100 / L40S, Morpheus, RAPIDSHuntra
Correlation graphLinks entities and anomalies into temporal attack graphs.RAPIDS cuGraphVeridex
Investigation agentPlans pivots, retrieves threat intel and writes the incident narrative.NeMo, TensorRT-LLMVeridex
Response enginePredicts blast radius and runs reversible containment through integrations.A100 / L40S, cuGraphQuellix
Edge sensorRuns detection on site for OT, IoT and air-gapped networks.Jetson Orin, IGX, BlueFieldSentix
SimulationEmulates adversaries and generates labelled synthetic telemetry.DGX / HGX, OmniverseSynthra
Console and APIIncident timelines, audit log, analyst feedback and access control.Cloud or your VPCPlatform

The flywheel

It hunts better the longer it runs

  1. Baselines deepen

    Each tenant's model of normal sharpens with every week of telemetry.

  2. Analysts give verdicts

    Confirmed and dismissed incidents become labelled training data.

  3. Synthra fills gaps

    Simulated techniques cover attacks your environment hasn't seen yet.

  4. Detection improves

    Anonymised behaviour patterns strengthen detection for every opted-in tenant.

Deployment

Runs where your data has to stay

Managed

Lairhunt Cloud

We run the GPU pipelines and console. The fastest way to start, with elastic capacity for telemetry spikes.

  • Dedicated tenant per customer
  • US or EU data region

Hybrid

Your VPC

Pipelines and data stay in your cloud account. We operate the software through a scoped role.

  • AWS, Azure or Google Cloud
  • Bring your own GPU capacity

On-prem

Your data centre

Everything on your hardware for environments that can't send telemetry out, delivered with NVIDIA AI Enterprise.

  • Air-gapped install option
  • Offline licence

Console

What your analysts work in

Queue

Incident queue

Incidents ranked by calibrated severity and confidence, not by alert count.

Timeline

Attack narrative

Each incident tells its story step by step, linked to the raw events behind it.

Graph

Entity graph

See which hosts, identities and workloads are involved and how they connect.

Hunt

Analyst queries

Ask questions of full telemetry in plain language or in query syntax.

Respond

Actions and rollback

Approve, run or reverse containment actions from the incident.

Feedback

Verdicts that train

Mark incidents as confirmed or benign; the models learn from both.

Security

Your telemetry is your data

Telemetry, incidents and baselines stay in your tenant. Shared detection improvements use only derived behaviour patterns you opt in to share, never raw events.

  • Encryption in transit and at rest
  • Customer-managed keys on Hybrid and On-prem
  • Per-tenant isolation of compute and storage
  • SOC 2 Type II and ISO/IEC 27001 audits planned for the first year of operation

Data you control

Telemetry, baselines, incidents, actions

Retention, region and sharing are settings in the console, not contract negotiations.

Trust and security details

Walk the platform with a hunter

A 45-minute session on your telemetry sources, your SOC workflow and where Lairhunt would fit. Pre-launch, so we show example incidents rather than a live tenant.