n8n Self-hosted AI Starter Kit 🧩 Workflow Open source

What it is

The Self-hosted AI Starter Kit is n8n's official Docker Compose template for running AI workflows entirely on your own machine. It starts n8n together with Ollama for local language models, Qdrant as a vector store and PostgreSQL, preconfigured so they talk to each other. It is a quick way to build private content workflows such as summarising documents, drafting posts from notes, or chatting with your own archive without sending data to cloud APIs.

Key features

• Docker Compose bundle of n8n, Ollama, Qdrant and PostgreSQL
• Profiles for NVIDIA GPU, AMD GPU and CPU-only setups
• Includes a demo AI agent workflow to start from
• Local models such as Llama and Qwen through Ollama
• Shared folder for files that workflows can read
• Maintained by the n8n team

How to install

Run git clone https://github.com/n8n-io/self-hosted-ai-starter-kit.git, enter the folder, copy the example .env, and start with docker compose --profile gpu-nvidia up (or the gpu-amd / cpu profile). On Apple Silicon the README suggests running Ollama natively.

Pricing and rights

Free, Apache-2.0 for the kit. n8n itself runs under its Sustainable Use License (free for internal use), and each local model has its own licence. No per-execution charges when self-hosted.

Who it is for

Privacy-minded creators, agencies and developers who want local AI automations and have Docker and a reasonable GPU.

Verdict

The easiest official route to a private AI automation stack. It is a starter, not a production setup: you still handle security, backups and updates, and local models trail frontier cloud models in writing quality.