Skip to main content
NodePad is built to be self-hosted. On the Enterprise plan, you deploy NodePad into your own cloud environment — your VPC, your servers, your network. NodePad’s application layer runs entirely inside your perimeter, and none of your conversation data touches NodePad’s infrastructure. This is the right deployment model for organizations that need data sovereignty, want to connect custom model infrastructure, or are subject to compliance requirements that preclude third-party data processing.

​
Why self-host

Data sovereignty

Conversation data, user activity, and audit logs remain inside your own environment. You control where data is stored and who can access it.

Custom model endpoints

Connect NodePad to your own inference infrastructure — Ollama, vLLM, Amazon Bedrock, Google Vertex AI, or any compatible endpoint.

Your own keys

Bring your own API keys for external model providers. NodePad never handles or stores your keys on its servers.

Compliance readiness

Keep all data flows inside your environment. Compliance reviews cover your infrastructure, not a shared SaaS platform.

​
Deployment options

The standard Enterprise deployment runs NodePad inside your VPC on your own cloud provider (AWS, Azure, GCP, or on-premises hardware). You provision the infrastructure; NodePad provides the application.This deployment connects to external model providers using your own API keys, so model inference calls go from your network directly to the provider — not through NodePad’s servers.
If your team also needs to connect local or self-hosted models, you can run Ollama or vLLM inside your VPC alongside NodePad and point NodePad’s model configuration at those endpoints.

​
Bring your own model endpoints

Self-hosted NodePad supports connecting to the following inference backends:
  • Ollama — run open-source models locally or on private GPU infrastructure
  • vLLM — high-throughput inference server for self-hosted models
  • Amazon Bedrock — AWS-managed model inference, accessed using your own AWS credentials
  • Google Vertex AI — Google Cloud model inference, accessed using your own GCP credentials
You can also use external model providers (OpenAI, Anthropic, etc.) with your own API keys, as long as your network allows egress to those endpoints.
If your compliance requirements prohibit any egress to external model providers, choose the air-gapped deployment option and run inference entirely inside your network.

​
What you control

In a self-hosted deployment, you own and operate:
  • The compute infrastructure NodePad runs on
  • All storage where conversation data and audit logs are written
  • Network routing and egress rules
  • Identity provider integration (SSO via SAML or OIDC)
  • Model endpoint configuration and API key management
NodePad does not phone home, send telemetry, or require an ongoing connection to NodePad-managed infrastructure once deployed.
Self-hosted deployments require your team to manage infrastructure provisioning, updates, and availability. The Enterprise team will guide you through the initial setup and provide documentation for ongoing operations.

​
Get started

Self-hosting is available to Enterprise customers. Contact the team to discuss your infrastructure environment, compliance requirements, and deployment timeline.

Contact the Enterprise team

The team will walk you through infrastructure requirements and the deployment process.