Test strategy support
Prepare coverage maps, regression focus areas, acceptance criteria checks, risk-prioritized test plans, and release-readiness questions for human review.
AI Testing Governance
Govern AI-assisted testing before it touches critical systems.
Scaled Agents helps teams design, govern, and review AI testing workers that can prepare test plans, generate candidate cases, route approvals, request scoped test execution, and preserve evidence without turning a draft testing workflow into production authority.
Offering Fit
A Scaled Agents offering for QA, delivery, engineering, risk, and platform teams.
AI-assisted testing can move quickly from helpful drafting to risky execution. This offering gives teams a governed path for deciding which AI testing workers may plan, generate, inspect, simulate, execute, report, or remediate work, and what must pause for human review.
Governed test generation
Create candidate manual, API, UI, data, integration, and workflow test cases from approved requirements, controls, tickets, and system context.
Controlled execution requests
Route proposed test execution through Passport scope, tool permissions, Toll Gates, Human Review, Runtime Permit checks, and Action Broker boundaries where execution is approved.
Evidence and audit trail
Record test intent, source context, approvals, run references, outcomes, defects, exceptions, missing evidence, and release-blocking decisions as review-preparation evidence.
Process Steps
From testing idea to governed testing worker.
Classify the testing use case. Identify the application, workflow, release path, test type, business owner, technical owner, QA owner, risk tier, data sensitivity, and production proximity.
Create the AI Testing Worker Blueprint. Define what the worker may draft, inspect, simulate, request, or report, and what it must never approve, alter, deploy, delete, or execute without review.
Prepare the Passport and tool boundary. Capture owner, purpose, approved scope, tool/API/connector candidates, data classes, prohibited actions, escalation route, and evidence requirements.
Attach Toll Gates and human review. Require review for sensitive data, production-like environments, external system calls, release-blocking findings, defect remediation, and any consequential action.
Route execution through controlled requests. If execution is in scope, require a scoped Runtime Permit and Action Broker decision before a test worker can invoke an approved tool or connector.
Preserve evidence and close the loop. Store reviewed plans, approvals, run references, outcomes, exceptions, defects, missing evidence, and remediation decisions in the activity and evidence trail.
This page is a public-preview planning artifact. It does not connect testing tools, run automated tests, approve production testing, authorize deployments, certify compliance, validate security posture, or replace accountable owner review.
Control Model
Testing workers need boundaries before they need tool access.
Allowed draft work
Summarize requirements, draft candidate tests, map coverage, compare expected and observed outcomes, prepare defect context, and identify missing evidence.
Restricted actions
Block self-approval, production changes, credential creation, deletion, deployment, customer-impacting actions, destructive tests, and unauthorized connector use.
Review evidence
Keep test source, owner decision, risk rationale, tool boundary, permit request, result reference, exception status, and remediation action visible for review.