Skip to content
Archively.ai

Security and governance

Who can do what, and the record of who did.

Access in Archively is explicit. Named roles hold named permissions. A review pipeline gates release. Every change is attributed, and published versions are kept.

Last reviewed July 2026

Roles
7 plus administrator
Workflow
9 stages
Authentication
SAML SSO · MFA
Audit
Field-level, attributed

Access and accountability

Permissions are named. Actions are logged.

01

Seven archival roles plus an administrator

Access is role based. Seven archival roles cover the stages of archival work, and an administrator manages the institution. People get the permissions their role needs, and nothing more.

7 ARCHIVAL ROLES + ADMINISTRATOR

02

Nine workflow stages

Records move through nine defined workflow stages before release. A review pipeline runs from AI review to quality review to rights clearance to publish. Release is gated, not automatic.

9 WORKFLOW STAGES · REVIEW-GATED RELEASE

03

Per-section permissions

Work on a record is organized by section, so each person's area of responsibility stays clear and their edits stay attributed to them.

PER-SECTION PERMISSIONS

04

Field-level audit trails

Every edit is recorded with the person who made it and the time it happened. Changes are attributed at the field level, so the history of a record stays legible.

FIELD-LEVEL AUDIT · ATTRIBUTED CHANGES

05

SAML SSO and MFA

Staff sign in through your identity provider with SAML single sign-on. Multi-factor authentication is supported for staff accounts.

SAML SSO · MFA

06

Published-version history

Published versions are frozen. When a record changes and is published again, the earlier version is kept, so you can see what was public and when.

PUBLISHED-VERSION HISTORY

How the record holds up

The AI proposes. A person decides.

The human-authority model is the spine of the platform. It is also part of its security story, because it means every published field was reviewed and attributed.

01

The AI proposes; archivists decide

AI output is an immutable snapshot. A person reviews it field by field and marks each field accepted, modified, or rejected, with a word-level view of the difference. The decision is human, and it is recorded.

IMMUTABLE AI PROVENANCE · FIELD-BY-FIELD REVIEW

02

Staff work and public access are separate

The staff workspace and the public portal are separate surfaces. Only published records appear on the public portal.

STAFF WORKSPACE · PUBLIC PORTAL

03

Each institution is separated

In the managed cloud, each institution runs in its own separated space, with its own catalog, media, users, and configuration.

PER-INSTITUTION SEPARATION

Put your security questionnaire into the evaluation.

A working setup using your records, not a sales presentation.