n8n and Ollama: a local LLM for private automation without the cloud

⚡ The gist in 5 seconds - n8n connects to a local language model via Ollama: instead of a cloud API, it makes an HTTP request to your own instance. - Ollama runs in a neighboring container, and n8n talks to it over the internal Docker network; setup takes about ten minutes. - The requirement is that the model runs locally and its endpoint is reachable from the n8n container; everything is computed on your own hardware. ### 🔍 What was found An article on Habr breaks down how to pair the open-source automation tool n8n with a local LLM via Ollama: both services run in Docker, the model spins up in a separate container, and n8n communicates with it using a plain HTTP Request node. No external API keys and no sending data to the cloud — the entire pipeline stays within your own infrastructure. The example task is generating summaries from incoming requests. The mechanics are simple: n8n sends a POST request to the local Ollama endpoint (by default, port 11434) with JSON like {"model": "<model-name>", "prompt": "<text>", "stream": false} and receives the response within the same workflow. From there, the result can be immediately parsed into fields, written to a database, or sent to a messenger — the automation logic remains on the n8n side. ### 💡 Why it matters For teams that don't want to route internal requests, correspondence, and documents through third-party APIs, the n8n + Ollama combo offers a workable compromise: familiar workflows with branching, retries, and integrations — but with the model running on your own server. The practical cost of this approach is hardware requirements and the quality of the local model, which depends on exactly what you've spun up in the container.