What they bring
A defined capability, tool manifest, data requirements, evaluation method, permission model, and failure controls.
PARTICIPANT 24 / 24
AI agent and automation builders
AI builders contribute bounded automation whose tools, data, permissions, logs, and escalation paths remain under human-governed controls.
ServiceStandard review
Typical participants
Agent builders, workflow-automation teams, AI operations specialists, model providers, and API services.
Agent builder · workflow automation · model/API
WCN role map
- Participant type
- ai_agent_builder
- Primary role
- Service
- Supporting roles
- Project · Track
- Review path
- Standard review
- Phase 1 path
- Eligible to apply as a horizontal capability, subject to agent-access review.
PARTICIPATION FIT
What this participant brings—and what they may seek.
What they may seek
Scoped workflows, authorized data, testable acceptance criteria, safe sandboxing, human review, and attributable run history.
AI + HUMAN CONTROL
AI may prepare. People retain decision authority.
Agent Workflow Copilot
The workbench describes a bounded assistance pattern for intake, structure, checks, and drafts. It is not proof that a public AI product or integration is live.
Mandatory Human Gate
Tool permissions, private data, external side effects, production access, approval, and every irreversible action.
PUBLIC BOUNDARY
Participation is reviewed, not automatic.
Agents may read or draft only within granted scope; they cannot approve their own output or create formal external effects.
- Eligible to apply as a horizontal capability, subject to agent-access review.
- Tool permissions, private data, external side effects, production access, approval, and every irreversible action.
- These tags name evidence categories for future reviewed work. They do not create public Proof, acceptance, settlement, certification, or endorsement.
Evidence vocabulary: service_proof · technical_proof
Service