Security
AI you can audit, line by line.
Tenant isolation, role scope, agent runs, and human overrides each leave a trail you can read.
Controls
Workflow walkthroughEvery control leaves a record.
Access, tenant isolation and audit evidence are separate controls, each carrying its own current status.
Trust
Control record
What arrives
Access policy
What you review
Audit evidence
Tenant isolation enforced
Source and state visibleWhat a governed run emits
Controls you can watch fire.
Isolation, role scope, agent execution, and human overrides each produce a line. The trace below is an illustrative run, not a customer record.
- 09:14:02tenant.scoperesolve org → workspaceok
What control means here
Nothing leaves the desk unsealed.
Each rule is a physical check. When one fires it stamps the record with the rule, the time, and the person or agent behind it — and the stamped copy is what you show an auditor.
- 01
RuleFlywheel
Each control is a fixed die. Change it and the change is versioned, not improvised.
- 02
CheckRam
The ram only travels when the record meets the die. A miss leaves the record untouched.
- 03
StampDie bed
A pass writes rule id, timestamp, actor, and the input hash onto the record itself.
- 04
KeepTray
Stamped copies collect in a write-once tray. You can read them; you cannot rewrite them.
Four controls, each with its own evidence.
Implemented, in audit, or planned — stated per control, with the source beside it.
Tenant isolation
Database-level row policies are the public proof point; implementation evidence is listed in source labels and reviewable artifacts.
Evidence bundles
Agent decisions, provenance, and human overrides are designed to be exportable for review and audit.
Access control
Enterprise identity is handled through organization-scoped authentication and role-aware API access with explicit boundaries.
Deployment control
Sovereign evaluation patterns are separated on the sovereign page for infrastructure review and deployment comparison.
Example data only.
Everything you see on this site is illustrative. No customer names, no policyholder data, no production records.
Design-partner preview only
Screenshots, reels, and animated cards use sanitized example content; no customer data, policyholder names, or production identifiers are displayed.
Privacy-safe telemetry
Public-site telemetry stays first-party and event-level only; it does not capture document contents, extracted fields, or form inputs.
Reduced-motion fallbacks
Motion surfaces degrade to static evidence cards when the browser requests reduced motion, so the proof remains readable and accessible.
First-party assets and scripts
The marketing site does not rely on third-party embeds or trackers, and its CSP constrains scripts, media, frames, and workers to approved origins only.
We do not imply production access, hidden workflows, or real policyholder data anywhere on this site.
Certification, stated plainly.
Row-level isolation
Tenant access is enforced at the database boundary, not only in application filters.
Database row-level security policies
Audit evidence
Per-submission evidence bundles keep agent and human decisions reviewable.
Audit evidence store
SOC 2
Public copy must state audit status rather than certification completion.
company-provided status
ISO 27001
Information security certification remains roadmap until verified.
roadmap status only
Review security posture with ReRao.
Bring a real submission packet into a scoped design-partner evaluation for broker intake, placement handoffs, operations, or sovereign deployment.
Evaluation requested
Review-ready