Trust and governance

Orientis is the AI division of Domus Solis LLC. This page is the public description of how AI features handle data.

No training on customer content

Text and files you submit so a feature can answer are used to produce that answer and to record the cost. They are not used to train models.

Provider

Inference runs on Amazon Bedrock, in our AWS account. The model can change. The product is billed in AI Units, not by a provider's retail token price.

Retention

We keep the accounting record of a call: product, account, model, units, latency, and cost. We do not promise to store the full prompt after the response is returned. Support mail you send us is kept to handle the request.

People

A person at Domus Solis may read a prompt when you ask us to, when we are fixing a failure you reported, or when we have to for a legal request. That access is not the normal path of a call.

Caps

Each plan has a unit balance and a maximum units per request. A call that would pass the balance stops. Paid plans can opt into overage with a monthly ceiling. The numbers on the pricing page are the draft until real traffic replaces them.

PII

A redaction layer that runs on every call is planned. It is not switched on for customers yet. Do not send secrets or card numbers to an AI feature.