---
title: "Self-Hosting - Documentation"
description: "Run AI for Database in your own infrastructure. Docker, Kubernetes, and bare-metal deployment guides."
canonical: "https://www.aifordatabase.com/docs/self-hosting/"
---

# Self-Hosting

advanced 8 min read

## Self-Hosting

Self-hosting AI for Database gives you complete control over your data. Nothing leaves your network -- database connections, queries, and results all stay within your infrastructure.

### Why Self-Host?

*   **Data sovereignty** -- no data leaves your network, ever.
*   **Compliance** -- meet strict regulatory requirements (HIPAA, PCI-DSS, government).
*   **Network simplicity** -- no need for SSH tunnels or IP allowlisting since everything is internal.
*   **Custom AI models** -- use your own LLM deployment or private API keys.

### Deployment Options

AI for Database is distributed as a Docker image and can be deployed on any infrastructure.

#### Docker Compose (Simplest)

```yaml
version: '3.8'
services:
  aifordb:
    image: aifordatabase/server:latest
    ports:
      - "3000:3000"
    environment:
      - DATABASE_URL=postgresql://user:pass@localhost/aifordb
      - OPENAI_API_KEY=sk-your-key
      - ENCRYPTION_KEY=your-32-char-encryption-key
      - JWT_SECRET=your-jwt-secret
    volumes:
      - aifordb_data:/app/data

volumes:
  aifordb_data:
```

Run with `docker compose up -d` and access the UI at `http://localhost:3000`.

#### Kubernetes

We provide Helm charts for production Kubernetes deployments:

```bash
helm repo add aifordb https://charts.aifordatabase.com
helm install aifordb aifordb/aifordatabase \
  --set database.url="postgresql://user:pass@db:5432/aifordb" \
  --set ai.apiKey="sk-your-key" \
  --set encryption.key="your-32-char-key"
```

The Helm chart includes configuration for replicas, resource limits, ingress, persistent volumes, and health checks.

### System Requirements

**Minimum (small team, < 10 users):**

*   2 CPU cores
*   4 GB RAM
*   20 GB disk

**Recommended (production, 10-100 users):**

*   4 CPU cores
*   8 GB RAM
*   50 GB SSD

**AI for Database requires a metadata database** (PostgreSQL 14+) to store configuration, user accounts, and query history. This is separate from the databases you connect as data sources.

### AI Model Configuration

Self-hosted deployments support multiple AI backends:

*   **OpenAI API** -- set `OPENAI_API_KEY` environment variable
*   **Azure OpenAI** -- set `AZURE_OPENAI_ENDPOINT` and `AZURE_OPENAI_KEY`
*   **Anthropic Claude** -- set `ANTHROPIC_API_KEY`
*   **Local models** -- point to any OpenAI-compatible API endpoint (e.g., Ollama, vLLM)

### Updates

Self-hosted instances receive updates through new Docker image tags. We publish stable releases monthly and patch releases as needed. Enable automatic update checks in the admin settings to receive notifications.

### Support

Self-hosted customers on Enterprise plans receive dedicated support including deployment assistance, upgrade guidance, and priority issue resolution. Community support is available on our Discord for open-source users.

Source: https://www.aifordatabase.com/docs/self-hosting/