EM-AI-01 · Reviewable research profile · 0.1.0

Enterprise AI Agent and Execution

A declarative Enterprise adoption profile over WM-AI-001, WM-AI-002, WM-AI-005, WM-AI-004. It creates no new model or runtime identity.

Verdict: PROFILE

Claude and Grok independently reviewed the boundary. One frozen no-tools audit and deterministic remediation are preserved. The profile contains 20 portable constraints and keeps base-model mastership external.

profile.json · fixtures.json · README.md

Independent evidence

Claude study · Grok study · Grok manifest · Provider comparison · Frozen audit

Limits

Endpoint and tool-service references require reconciliation with WM-SFT-018, WM-SFT-010, WM-SFT-003 and service boundaries.
The AI-01, AI-03, AI-04 and AI-07 field-level crosswalk is incomplete.
Base models retain source-pin, ISO, deprecated MCP and development-telemetry holds.
The external grant, approval and runtime endpoint masters must prove the identity and lifecycle conditions required to avoid ToolBinding and InferenceEndpoint masters.
Per-tool scope, serving-contract and dual-digest extensions require multi-profile validation.
One frozen independent audit remains pending.

This profile is not canonical, installable or a standards-conformance claim.