Trust
Security you can verify, claims we can back
This page describes controls that are implemented in the product today. Scholar XI does not hold SOC 2, ISO 27001, HIPAA or FedRAMP attestations, and we do not display badges for them.
Implemented controls
Tenant isolation
AvailableDatabase row-level security, scoped storage paths, scoped caches and per-tenant queue leases.
Server-side authorization
AvailableThe client never decides access. Every server function re-checks the caller's role.
Server-held credentials
AvailableProvider and connector credentials never reach the browser.
Data classification
AvailableWorkspaces inherit a classification; confidential scopes block external tool calls.
Encryption in transit
AvailableAll traffic to Scholar XI and to connected services uses TLS.
Audit logging
AvailableAdministrative and approval events are recorded with actor and timestamp.
How isolation works
Every tenant-owned row carries its organization and workspace. Access is enforced by row-level security policies in the database, so a query that escapes an application check still returns nothing. Storage objects live under tenant-scoped paths and are served through short-lived signed URLs.
Semantic caches, background job queues and memory stores are keyed by tenant scope. Entries written under one scope cannot be read under another, and legacy or forged keys are rejected rather than tolerated.
Workspaces carry a data classification that projects and conversations inherit. When a scope is classified confidential, XI is blocked from calling external tools with that content.
Provider and connector credentials are held server-side only. They are never sent to the browser, and internal model infrastructure is not exposed in customer-facing responses.
Running a security review?
Send us your questionnaire and we will answer it against what is actually implemented.