Your data stays yours.
Where your data lives, who can see it, which models touch it and on what terms, and what happens when we stop. The short version: it stays in your environment, and you can see everything the system did.
The short version
It stays in your environment. We read through the interfaces your systems already have.
A named person on your team approves every output, and every field shows its source.
No model is trained on your data. Hosted models run under written retention limits, or inside your network.
When we stop, access ends, and the code, prompts and evaluations are already yours.
In detail
Where it lives.
Your documents and records stay in your systems and your accounts. The system we build runs in your cloud or on your machines, with your keys. We do not copy your data into a system of ours, and there is no infrnce platform holding it.
Who sees it.
The named engineer, and the people you name. Access is granted by you, in your own systems, and you can revoke it at any time. We sign a confidentiality agreement before discovery, and we do not name you as a client without your written permission.
Which models, on what terms.
Documents are read by models you approve. Where a hosted model is used, it runs through a private endpoint under a contract that bars training on your data and limits how long it is held. Where that is not acceptable, open models run entirely inside your network. The choice is written into the scope.
What the system is allowed to write.
Nothing enters your system of record on the model's word alone. The system writes to a review queue or a sheet your team controls, and a named person approves each item. Every field keeps a link to the document it came from.
The record of what happened.
Every read, every proposal and every approval is logged in your environment, in a log you own. If a question comes up later, the answer is in your records, not ours.
What leaves your environment.
Only what the scope lists, in writing, and only to the services it names. If nothing is listed, nothing leaves. Any change needs your written agreement first.
When we stop.
Access is revoked. Code, prompts, evaluations and any fine-tuned weights are already in your repositories and accounts.
Asking us about this.
Email contact@infrnce.com. We answer questions about data handling before any engagement, in writing.
Put an engineer inside your operation.
Tell us about the work your team still does by hand.
Book a call