Under the hood

The platform’s current status and incident history, updated independently of this site.

1232 automated end-to-end tests across 102 suites, last run October 1, 2026 (SYSRUN-20261001-KGXR), health score 100/100.

A tenant can run the platform’s own test suite against their own account from inside the app and see the results — not a number we publish on their behalf.

Issued certificates and reports are hash-locked at issue. Anyone holding a certificate number can verify it independently.

Every AI agent in the codebase is either registered — with an owner, a stated purpose and whether it writes to a record — or explicitly assessed out of scope with a reason. 105 are registered today, 52 are assessed out of scope, and a build fails if a new agent appears in neither list. Narrative output that reaches an issued document is released only after a human approves it.

A continuous compliance-readiness view computed from a tenant’s own live records — open CAPAs, equipment due for calibration, training and proficiency testing status, deviations — not a number anyone enters by hand.

Single sign-on via OpenID Connect (Microsoft Entra ID, Google Workspace, Okta), two-factor authentication, and password re-authentication at the moment of signing on 33 sensitive actions — a logged-in session alone is never enough to sign or issue.

A library of the regulatory standards and frameworks referenced across the platform, with a service that flags when one may have moved and needs a human look.

Tenant isolation, encryption, access control, the audit trail and procurement information — controls aligned to SOC 2 and ISO 27001, audit in planning.

Load-test runs produce a SHA-256 hash-locked record of what was measured — including the shared-pool vs. dedicated-database distinction a published figure has to state, after an earlier run’s ambiguity on exactly that point was found and fixed.