Read-scope & consent.
You decide what the AI may observe, source by source. Connect only what you want, revoke any time, and see exactly which signals informed a proposal.
Security & governance
Human-in-command by design.
MOS uses AI to keep your source of truth current — but the AI only ever proposes. Nothing enters the record without a human, and you control exactly what it's allowed to observe.
The core guarantee
The ratification gate.
Every change the AI suggests is a proposal, not an edit. A human with the right permission reviews each one — with the evidence attached — and approves or dismisses it. Nothing is ever written to your source of truth automatically. This is the difference between a tool that drifts and one you can stand behind.
Policy P-07 — Status
Current: Status: Current
Proposed: Status: Superseded (proposed)
A newer policy was approved and circulated last week.
Permissions & ownership.
Every artifact has a named owner. People and agents see only what they're entitled to — no more, no less — and only owners can ratify changes to what they own.
Governance & audit log.
Every proposal, approval, and dismissal is recorded. The changelog is a complete, reviewable history of how your source of truth came to say what it says.
Not surveillance.
MOS surfaces structural misalignment — where the written record has drifted from reality. It is not a tool for monitoring individuals or scoring performance, and it never will be.
Data handling.
[placeholder: data residency, encryption in transit/at rest, retention — fill with verified specifics]
Enterprise controls.
[placeholder: SSO, SCIM, roles, hosting, uptime, backups — fill with verified specifics]
Compliance.
[placeholder: certifications pending — do not display any badge until verified]