Skip to main content
L3gitUI · API · MOBILE TEST AUTOMATION
[ PRODUCT ][ PRICING ][ DOCS ][ VERIFIED ][ REGULATED ][ COMPARE ][ ACCOUNT ] Start Free Trial

Home › How-to guides › How to run AI-assisted test automation fully offline with Ollama

/// AI · air-gapped

How to run AI-assisted test automation fully offline with Ollama

For networks that cannot send code or page content to a cloud model. The AI runs on your own PC; no key, no account, nothing leaves the machine.

About 15 minutes · step by step, no experience needed

  1. Install Ollama from ollama.com (no account needed — skip the sign-in). Then open a terminal and run ollama pull llama3.1:8b (about 5 GB, one time).
  2. In L3git, open Tools → Settings and integrations.
  3. Set Provider to Local model via Ollama. Leave the URL as http://localhost:11434 and the model as llama3.1:8b (or whichever you pulled).
  4. Click Test — it should say the model is installed — then Save.
Ollama can also run on an internal server: put that machine's address in the URL box. An 8B model is fine for authoring and repair; if its drafts feel weak, ollama pull qwen2.5:14b and switch the model name. The AI is used only while you author; generated tests never call a model.

Done! The same AI buttons, with zero outbound traffic.

L3git for air-gapped teams — Every edition runs offline; the AI helper runs on a model you host. Nothing leaves the network.

Start the 14-day free trial   About L3git for air-gapped teams

All guides: /guides · Deeper write-ups: Resources · Still stuck? Email corwinhill@l3git.com — a real person replies.