How it works
The trust architecture behind the chat box
Your prompt and files run through a private inference layer where models are accessed without retaining your data for training. Your files live in an isolated private space used only to answer you.
Step by step
- 1
You
Your prompt and any files go to the app — nothing more than you choose to share.
- 2
Private inference layer
Models are accessed so your prompts and data are not retained for training.
- 3
Your private space
Your files and history live in an isolated space, used only to answer you.
- 4
Response
Encrypted in transit and at rest, regionally hosted — and you hold the controls.
Guarantees, built in
No training on your data
Encrypted in transit & at rest
Regional hosting; you control retention & deletion