Skip to main content
NodePad’s air-gapped deployment mode is designed for environments where data cannot leave the network under any circumstances. In this configuration, NodePad operates entirely inside your network perimeter: the application runs on your infrastructure, models run on your own hardware or private GPU clusters, and no outbound connections are made — not to NodePad’s servers, not to external model providers, not to anyone. This is the most restrictive deployment option NodePad offers, and it is designed to meet the requirements of the most security-sensitive organizations in the world.

​
How it works

In a standard self-hosted deployment, NodePad may make outbound calls to external model providers using your API keys. In an air-gapped deployment, that path is eliminated entirely. All inference happens on model servers you run inside your own network.
1

NodePad runs inside your network

The NodePad application is deployed on infrastructure you control, with no outbound internet access permitted at the network level.
2

Models run locally

You run a local inference server — Ollama, vLLM, or a compatible endpoint — on hardware inside your perimeter. NodePad points its model requests at this server.
3

Zero data leaves your network

Conversations, prompts, completions, user data, and audit logs all stay inside your environment. Nothing is transmitted externally.

​
Supported local inference backends

NodePad is compatible with the following self-hosted inference solutions for air-gapped deployments:

Ollama

Run open-source models locally with a lightweight inference server. Suitable for smaller teams or organizations with limited GPU resources.

vLLM

High-throughput, production-grade inference server for organizations running larger models at scale on private GPU infrastructure.
Amazon Bedrock and Google Vertex AI require outbound connectivity to AWS and GCP respectively and are not compatible with fully air-gapped deployments. For air-gapped environments, use Ollama or vLLM with locally hosted model weights.

​
Who this is for

Air-gapped deployment is the right choice for organizations with the strictest data egress controls:
  • Defense and intelligence agencies operating under classified network policies
  • Government contractors subject to FedRAMP High, IL4, IL5, or IL6 requirements
  • Financial institutions with internal policies prohibiting external data transmission
  • Healthcare organizations handling sensitive patient data under HIPAA or equivalent regulations
  • Critical infrastructure operators with network segmentation requirements
  • Any team where the answer to “can this data leave our network?” is always no

​
What you own

In an air-gapped deployment, your organization operates and is responsible for:
  • The network environment and its egress restrictions
  • All compute infrastructure running NodePad and your model servers
  • Model weights — you source, store, and manage the models you run
  • All storage for conversations, user data, and audit logs
  • Identity provider integration (no external IdP connections are required — you can run your IdP inside the same air-gapped environment)
Air-gapped deployments require more infrastructure planning than standard self-hosted deployments. You are responsible for sourcing model weights, provisioning GPU hardware, and managing the full stack inside your perimeter. The NodePad Enterprise team will help you plan and validate your architecture before deployment.

​
Data handling

In an air-gapped deployment:
  • No conversation data leaves your network. Prompts and completions are processed by your local inference server and stored in your environment.
  • No telemetry is sent. NodePad does not require any outbound connectivity once deployed.
  • Audit logs stay in your infrastructure. You own the audit trail and can export it to any system inside your environment.
  • NodePad does not train on your data. This is true for all NodePad deployments — and in an air-gapped environment, there is no technical path for data to reach NodePad’s systems anyway.

​
Get started

Air-gapped deployment is available to Enterprise customers. The team will work with you to assess your environment, validate that your infrastructure meets the requirements, and guide you through the deployment process.

Contact the Enterprise team

Describe your network environment and data egress requirements. The team will help you design the right deployment architecture.