Identity
Which AI worker is acting, under which owner, with which Passport and lifecycle state?
AI Assurance Framework
Organize AI worker assurance around identity, authority, runtime controls, and evidence.
Scaled Agents frames AI assurance as an operating-record problem: who owns the AI worker, what it may do, what control path applies, what evidence exists, and when a request should fail closed or route to human review.
This page is public orientation for enterprise buyers, reviewers, and procurement teams. It does not provide certification, legal advice, audit conclusions, or a claim that Scaled Agents itself is the object of a deployed-agent assessment.
Context
Use this page to understand how Scaled Agents fits assurance programs.
Enterprises increasingly need a common way to explain AI worker ownership, oversight, and runtime controls across internal review, customer diligence, and independent assessment work.
Scaled Agents provides the control-plane records and customer-safe evidence model around that work. The customer still owns deployment scope, production decisions, security controls, and any formal interaction with external assessors.
- Framework-aware mapping, not framework conformance claims
- Customer-managed evidence organization, not outsourced accountability
- Independent review may apply to a deployed AI worker system, not this public website
Assurance Pillars
Four operating questions define the assurance posture.
The product model is useful when assurance has to connect governance intent to runtime behavior and reviewable records.
Authority
What purpose, boundaries, tools, data classes, and prohibited actions have been approved?
Enforcement
Which Toll Gates, Runtime Permit checks, and Action Broker boundaries decide whether a request proceeds?
Evidence
Which Human Review records, Stamps, workflow events, and audit exports support later review?
Claim boundary
Scaled Agents does not certify compliance, validate security posture, provide an audit opinion, or approve production use. This page explains the control model that can support customer-managed review preparation.
Control Mapping
Map assurance needs to Scaled Agents capabilities.
This matrix shows the public operating fit between enterprise assurance expectations and the core product records.
| Assurance need | Current Scaled Agents capability | Fit | Recommended follow-through |
|---|---|---|---|
| Named accountability | Passport owner, sponsor, reviewer, escalation path, and lifecycle state. | Strong | Surface assurance profile fields in customer-facing Passport review workflows. |
| Runtime boundary enforcement | Toll Gate, Runtime Permit, Action Broker, and fail-closed decision path. | Strong | Keep customer deployment evidence separate from public product context. |
| Evidence reconstruction | Stamps, evidence records, workflow events, and audit export package planning. | Partial | Finalize customer-safe assurance export composition over existing records. |
| External review preparation | Framework mapping, role separation, shared responsibility language, and control-plane summaries. | Partial | Pair product records with customer environment evidence and independent reviewer input. |
Product Records
These are the records that matter for review preparation.
Identity, owner, purpose, scope, risk, lifecycle, and evidence linkage.
Policy and control checkpoints for data, autonomy, tools, and consequential action.
Accountable reviewer decisions, exceptions, escalations, and remediation instructions.
Customer-safe summary path over existing evidence objects for later reconstruction.