Your AI. Your key.
Your secrets never leave the building.
Andorg is the privacy middleman for AI. Bring your own provider key. Change one line — the base URL — and every request is masked on our machine before it leaves, forwarded on your key, and restored on the way back. We keep nothing.
No free plan. 14-day money-back guarantee.
Masked in our house
Hardcoded rules plus a local AI model find every private detail on our own CPU. No cloud model ever sees your raw text.
Your key, your bill
We never resell tokens. Your request goes to your provider on your key. We are the privacy layer, nothing more.
We keep nothing
Real values and your provider key live in memory for one request, then they are gone. Not in any database, log, or backup.
Two plans. One obvious choice.
Core
$29/mo
The privacy gate for chat.
Scale
$99/mo
Everything in Core plus attachments, 1,000 pages included.
Questions, answered
What is a private AI API?
A private AI API is a gateway that removes personal and sensitive data from your prompts before they reach an AI model, then restores it in the answer. Andorg does this on its own machines, so no cloud model ever sees your raw text.
How do I stop PII reaching an LLM?
Point your SDK's base URL at Andorg and send your provider key in a header. Andorg masks every detected value locally, forwards the clean text on your key, and restores the real values in the response. Nothing is stored.
Do you store my provider key or my data?
No. Your provider key and every real value live in memory for a single request and are destroyed when the answer is sent — never in a database, log, or backup.
Which AI providers work?
OpenAI, Anthropic, Google, OpenRouter, and any OpenAI-compatible endpoint. You always speak the OpenAI format, in and out.