Step-by-Step Guide: 100% Free Local AI Automation with n8n + Ollama (Multilingual)

Getting Started with n8n Docs & Tutorials
, , ,
1

Hi n8n community! :waving_hand:

I wanted to share a project I’ve been working on to help users run AI automations without the recurring costs of OpenAI or other paid APIs.

I’ve created a comprehensive guide on how to connect n8n with Ollama (Llama 3, Mistral, Phi-3) to run everything locally on your own hardware.

:glowing_star: Why this guide?

  • Privacy-First: Your data never leaves your machine.
  • Zero Cost: No API keys or subscriptions needed.
  • Multilingual Support: I’ve written the guide in English, বাংলা (Bengali), हिन्दी (Hindi), Español, Français, and العربية.
  • Ready-to-use Workflow: I’ve included a downloadable .json template so you can import and test it instantly.

:open_book: You can find the full guide here:

:backhand_index_pointing_right: GitHub: n8n-local-llm-ollama-guide

:rocket: What’s inside?

  1. Simple setup for Ollama on Windows/Mac/Linux.
  2. How to expose Ollama’s API for n8n.
  3. n8n HTTP Request node configuration.
  4. Hardware optimization tips for 4GB, 8GB, and 16GB+ RAM.

I’d love to hear your thoughts! If you have built any cool local AI workflows, please share them here. If you find the guide helpful, feel free to give it a :star: on GitHub!

ollama #LocalLLM #SelfHosted ai #Automation

3 Likes
2

It’s a good start

1 Like
3

The multilingual packaging is a nice touch - that’s an often-overlooked barrier for community adoption. One thing worth adding to the guide is a note about setting OLLAMA_HOST=0.0.0.0 when n8n is running in Docker, since by default Ollama only binds to localhost and that catches a lot of people off guard.

1 Like
4

This is great, thanks for putting it together. Running everything locally on Ollama is honestly underrated for anyone testing workflows before committing to paid API costs. Curious how it performs on larger multi step automations, do you notice much of a speed difference between Llama 3 and Mistral for real time chains, or does hardware end up being the bigger bottleneck either way. The multilingual writeup is a nice touch too, most guides skip that entirely.