Privacy
Personal details are shielded first. Names, numbers, and sensitive information are masked before processing begins - and you can see it happen, right in the interface.
Only the context your request actually needs is used. Nothing more is touched, and nothing is stored beyond your own account.
The answer comes back to you, and only you. It's never kept for another purpose, because there is no other purpose.
Your conversations are encrypted in transit and at rest. What you discuss with Dvina stays between you and your context - not on someone's screen, and not in someone's logs.
Dvina detects personal details automatically and masks them before anything else happens. And it doesn't hide the fact - masked details appear masked, right where you're working, so trust never has to be taken on faith.
Everything Dvina remembers is yours to inspect, correct, or erase - gone from the memory, not just the screen. Your context works for you, on your terms.
Cloud, private cloud, or fully on-premise. Dvina is LLM-agnostic and can run in a completely closed loop - so for teams where data can't leave the building, it never does.
Not a toggle
No exceptions
Applies to everything
You. Conversations are encrypted, personal details are masked, and access stays with your account. Dvina's job is to understand your context - not to expose it.
Models receive only what a request needs, with personal details already masked - and nothing they process is retained or used for training. The same rule applies to every model, every provider.
Yes. Remove single memories, whole topics, or your entire context. Deleted means deleted - from the product and from the memory behind it.
That's exactly who on-premise deployment is for. Dvina can run inside your walls, in a closed loop, with your choice of models - nothing leaves your infrastructure.
We only collect analytics essential to ensuring smooth operation of our services.