On-Premise Deployment

Your data, your infrastructure — full control inside your network

On-Premise

Built for Industrial Control Networks

Sentinel runs entirely inside your facility. Production images, trained models, and inspection results never leave your network. Direct GPU inference keeps inspection latency low enough for in-line decisions.

  • Data SovereigntyAll images and models stay within your network perimeter
  • Ultra-Low LatencyDirect GPU inference with no network roundtrip
  • Air-Gap SupportRuns fully offline — no internet connection required
  • Full ControlManage updates, backups, and configuration on your terms

Hardware Options

Run Sentinel on hardware you already own, or let us deliver a ready-to-go appliance

Bring Your Own Hardware

Use the Servers You Already Have

Install Sentinel on your existing infrastructure. We support a wide range of NVIDIA GPUs and standard server configurations — no special hardware required.

  • No Hardware PurchaseReuse capacity from your existing data center or edge servers
  • Standard ComponentsCompatible with NVIDIA GPUs (RTX 3060 and up) on Linux or Windows
  • We Validate the SetupPre-install checks confirm your hardware meets performance targets
  • Software-Only SupportHardware stays under your existing IT contracts

Larix-Provided Hardware

Turnkey Inspection Appliance

We deliver a pre-configured GPU server with Sentinel installed, tuned, and tested for your inspection workload. Plug it into your network and start training.

  • Ready to DeployHardware sized to your throughput and arrives pre-installed
  • Optimized GPU StackNVIDIA GPU, CUDA, and drivers all pre-tuned for Sentinel
  • Single Point of ContactOne vendor for hardware and software — no finger-pointing
  • Lifecycle SupportHardware warranty, replacements, and upgrade path included

Getting Started

1

Assessment

We analyze your inspection requirements, image types, defect categories, and production environment to recommend the optimal setup.

2

Pilot

Deploy Sentinel with your real production images. Build and validate initial models, measure detection accuracy, and tune thresholds.

3

Rollout

Integrate Sentinel into your production workflow. Connect to your MES/SCADA systems, configure automated inspections, and train your team.

4

Support

Ongoing technical support, model optimization assistance, and platform updates to keep your inspection system running at peak performance.

Technical Requirements

Specs to check if you're bringing your own hardware. With a Larix-provided appliance, we size and configure everything for you.

GPU

NVIDIA GPU with CUDA support (RTX 3060+ recommended)

RAM

16 GB minimum, 32 GB recommended

Storage

SSD with 100 GB+ free space for models and data

OS

Linux (Ubuntu 20.04+) or Windows 10/11

Bring Your Own or Let Us Deliver?

Tell us about your infrastructure and inspection volume — we'll recommend the right hardware path and a complete deployment plan.

Talk to Our Team