Loading…
A plain-language summary of how Kriyava handles data, models, and confidentiality. Full legalese available on request.
We default to retrieval over fine-tuning. When fine-tuning is necessary, it happens in isolated workspaces with documented provenance — never pooled.
We never train on your content without an explicit, written agreement scoped to your tenant. Period.
For Sovereign tier, your data and model weights stay inside your VPC. We have no access except to logs you explicitly share for support.
Telemetry — performance, error, cost — is collected by default and used to improve the platform. You can opt out at the workspace level.
Data deletion: 30 days from request. Backups purged on the same cadence.
Subprocessors: list maintained at kriyava.ai/security. Updated 7 days before any change.
Questions? privacy@kriyava.ai — replies within 5 business days.