Run the AI on your own machine
Point AISiteVolt at Ollama, LM Studio, vLLM or LiteLLM. No API key, no per-word bill, and no prompt ever leaves your server.
No AI bill
A local model costs nothing to run. Generate a hundred drafts and the invoice is still zero.
Nothing leaves your server
Prompts and content go to an address on your own machine. Not to us, not to OpenAI, not to anyone.
Works offline
No internet, no problem. Useful on a laptop, on a plane, or behind a client's firewall.
Four steps
- 1Install Ollama and pull a model — one command: ollama pull llama3.2
- 2In WordPress, open AISiteVolt → Settings → API Keys
- 3Enter http://localhost:11434 and your model name, then press Test
- 4Pick “Local AI server” as the provider and write as usual
Any OpenAI-compatible server
If it speaks the OpenAI chat API, it works. These are the common ones:
localhost:11434localhost:1234localhost:8000your gatewayWhat it does not do
Local models handle text: articles, rewrites, product copy, FAQ sections, GEO fixes. Image generation, transcription and speech still need a cloud provider — and a small local model writes noticeably weaker long-form content than a frontier model. You can switch provider per task.
Built for the way agencies work
Build the site locally, generate the whole content set offline at no cost, then switch to a cloud model for the pages that matter most. The client's material never leaves your laptop while you work.
Read the setup guide