After experimenting with various AI and DevOps tools, I found the following architecture to be a reliable, scalable, and easy-to-manage setup for running AI applications at home or in a small office.
Ubuntu 24.04 LTS │ ├── Docker Engine ├── Coolify ├── Portainer ├── Watchtower ├── Duplicati (Backups) ├── Uptime Kuma │ ├── AI Services │ ├── Ollama │ ├── Open WebUI │ ├── ComfyUI │ ├── n8n │ ├── Qdrant │ ├── PostgreSQL │ ├── Redis │ └── MinIO │ ├── Reverse Proxy │ └── Caddy (or Traefik) │ └── Cloudflare Tunnel (Optional)
This combination provides an excellent balance between deployment automation, container management, security, monitoring, backups, and AI workload management.
| Component | Purpose |
|---|---|
| Coolify | Deploy and manage applications using Docker Compose or Git repositories. |
| Portainer | Provides an intuitive web interface for managing Docker containers, volumes, networks, and logs. |
| Watchtower | Automatically updates Docker containers that use published images. |
| Caddy / Nginx Proxy Manager | Handles reverse proxying, HTTPS certificates, and custom domains. |
| Uptime Kuma | Monitors the health and availability of all running services. |
| Duplicati | Creates scheduled backups of application data and configuration. |
Best for: Self-hosting AI applications, Docker Compose stacks, and web applications.
Best for: Managing Docker containers and troubleshooting running services.
Dockge is developed by the creator of Uptime Kuma and focuses on Docker Compose management.
CasaOS is designed primarily for home servers and media servers.
While CasaOS is excellent for home media servers, it is less suitable for AI development and production workloads compared to Coolify or Portainer.
This setup offers the perfect balance between deployment automation, operational visibility, monitoring, security, backups, and long-term maintainability.