selfhost-ai

Why we included this project

Most self-hosted AI setups start as a personal folder of hand-written docker-compose fragments that only the author understands. This project skips that whole stage: it's a tested Docker Compose template with an interactive installer that brings up n8n workflow automation, Ollama local models, Flowise agent builders, vector databases, and Open WebUI behind a Caddy reverse proxy with automatic HTTPS. The queue-mode n8n running on Redis and Postgres, health checks, and the bundled Grafana and Prometheus monitoring make it feel like a small production rollout rather than a homelab experiment. Because the pieces are mature open components everyone already knows, you end up with familiar wiring for a self-hosted Zapier or Make alternative and can swap in your own models or integrations. Homelab owners and small teams get a private automation workshop without hand-rolling the infrastructure.