Designed for teams who can't take chances with their documents.
Giving an AI tool read access to your company's knowledge is a real decision. This page documents exactly what Knorra does with your data, what it never does, and how that's enforced.
Eight commitments. Every one technically enforced.
Your documents, tickets, and messages are not used to train models, ours or anyone else's. Contractually documented in every DPA. Architecturally, retrieval-augmented inference reads your tenant's index per request; no batch export to any training pipeline exists.
Knorra requests the narrowest read scopes a connector offers. Write scopes are opt-in, per source, and can be revoked at any time from the workspace settings.
Findings are surfaced, never applied. No silent edits, no background agents acting on your data. When an action requires write access, you approve it on the finding card and we record the trail.
Anything sourced from a personal Google Drive or other individual-account connector is visible only to that person. Personal-source findings are never team-wide, ever.
Email is out of scope. We do not request the Gmail OAuth scope, we do not index it, and we will not add it to the product roadmap.
Every sub-processor we use is listed publicly, including the data category each one touches. Subscribers receive email notice at least 30 days before any change.
Knorra SaaS runs on EU/UK infrastructure. Your data is stored and processed there. Region is not customer-selectable in the standard SaaS product. Enterprise customers can choose a dedicated independent deployment with the region of their choice.
Data in transit uses TLS 1.3 with HSTS. Data at rest uses AES-256 with envelope encryption. Per-tenant keys are rotated automatically every 90 days. On-demand rotation is available.
Stored and processed on EU/UK infrastructure.
Knorra SaaS runs on EU/UK infrastructure. Indexing and embedding happen there. Model providers see individual retrieval-augmented requests, governed by zero-retention agreements.
Enterprise customers can choose a dedicated independent deployment with the region of their choice.
Every third party that helps run Knorra.
Sub-processors are the third-party services that help run Knorra, such as hosting, AI inference, and email. We disclose every one and notify you before any change.
Subscribers receive email notice at least 30 days before any sub-processor is added, replaced, or its data scope changes. Subscribe from the footer.
Independent, documented, current.
Report available under NDA as the window closes.
Gap assessment complete. Stage 1 scheduled.
Executive summary available under NDA.
DPA signed at sign-up. Custom terms for Business.
We'll send our SOC 2 report (under NDA), the latest pen-test summary, our sub-processor list, and our architecture overview ahead of the call.