answers from your own machine · nothing sent anywhere

🧠 The Local Mind

Run a model on your own device — an Ollama server, or any OpenAI-compatible local endpoint — and the Seeker hands it questions. The question and the answer never leave this machine: no box, no cloud, no harvest. If you have not started one, install Ollama (ollama.com) and pull a model (ollama run llama3).

Your endpoint

Endpoint (loopback only — remote URLs are refused) Model

Run it in this tab — no server, no setup

No Ollama? The Local Mind can still answer. It downloads a small model into this tab once, then every question runs here through WebGPU. The question and the answer never leave your machine — the only thing fetched is the model itself, from a type CDN, and that is a download in, not a question out.

Model (downloaded once into this browser)

Ask

The Local Mind is strictly opt-in. The endpoint lane only ever talks to localhost / 127.0.0.1 / ::1 — a remote URL is refused before a single byte leaves. The in-browser lane fetches only a model from cdn.jsdelivr.net and huggingface.co, then computes every answer in this tab: the question and the answer never leave your machine. The endpoint, model, and choice of lane live in this browser's localStorage — the box never sees them.