Self-hosted AI Package

What is Self-hosted AI Package?

Self-hosted AI Package is an open-source Docker Compose template that bootstraps a fully featured local AI and low-code development environment. It integrates Ollama for local LLMs, Open WebUI for chatting with n8n agents, Supabase for database and authentication, plus Flowise, Langfuse, SearXNG, and Caddy. It is designed for developers who want to build and run self-hosted AI workflows with complete data control.

How to use Self-hosted AI Package?

Clone the repository, set environment variables in a .env file (copying from .env.example), then run the start_services.py script with a profile flag for your hardware (e.g., --profile cpu for CPU-only, --profile gpu-nvidia for Nvidia GPUs). Access n8n at http://localhost:5678/ to set up your local instance and import workflows. For e-commerce features, optionally deploy MCP servers and import the included n8n workflow.

Key features of Self-hosted AI Package

• Self-hosted n8n with 400+ integrations and AI components
• Supabase for database, vector store, and authentication
• Ollama to run local LLMs
• Open WebUI interface for chatting with n8n agents
• Flowise no/low-code AI agent builder
• Langfuse LLM observability and monitoring
• SearXNG privacy-focused meta search engine
• Caddy managed HTTPS for custom domains
• Qdrant high-performance vector store
• Optional e-commerce MCP servers for Shopify and Pictorem
• Pre-built multi-agent n8n workflows for order fulfillment

Use cases of Self-hosted AI Package

• Build and run AI workflows locally without external API dependencies
• Automate e-commerce order fulfillment from Shopify to print-on-demand (Pictorem)
• Create custom RAG (retrieval-augmented generation) agents using local data
• Develop and test AI automations privately before deploying to production
• Integrate multiple AI services (LLMs, vector stores, search) in a single environment

FAQ from Self-hosted AI Package

What are the prerequisites?

Python, Git/GitHub Desktop, and Docker/Docker Desktop are required.

How do I run Self-hosted AI Package with a GPU?

Use python start_services.py --profile gpu-nvidia for Nvidia GPUs or --profile gpu-amd for AMD GPUs on Linux. For Mac/Apple Silicon, use --profile cpu or --profile none to run Ollama natively.

What is included in the package?

It includes n8n, Supabase, Ollama, Open WebUI, Flowise, Qdrant, SearXNG, Caddy, Langfuse, and optional e-commerce MCP servers for Shopify and Pictorem.

How do I deploy to the cloud?

Set up A records for subdomains, configure Caddy environment variables in .env, open required ports (e.g., 80, 443, 5678), then run the start script on a Linux cloud instance.

Where do I get help or report issues?

Join the Local AI community forum in the oTTomator Think Tank and check the GitHub Kanban board for feature requests and bug tracking.