Coolify gives you a Vercel-style deploy UI on any €5 VPS with Docker and Traefik doing the heavy lifting. For side projects and small SaaS, it saves you $20+/month and gives you full control — but you trade Vercel's edge network for single-VPS ops responsibility.
Ollama + Open-WebUI in Docker Compose is the boring, correct way to self-host a local LLM. Pass through your GPU with nvidia-container-toolkit (or ROCm for AMD), put Open-WebUI behind Traefik on Tailscale, and run Q4_K_M quantised models sized to your VRAM — 8B on 12GB cards, 70B on 24GB.
Full Docker Compose setup guide for Glance Dashboard v2 — RSS feeds, Uptime Kuma integration, server stats, and an honest Glance vs Homepage verdict.
Run a free, always-on phone LLM server using a Pixel 9 Pro, OlliteRT Gemma 4 E4B, and Tailscale — wired into n8n and Home Assistant for real homelab tasks.