/uae-ai-autonomy-tier

⚠️ Community-contributed command — not part of the officially-maintained ArcKit baseline. Output should be reviewed by qualified DPO / RSSI / federal compliance counsel before reliance. Citations to UAE Cabinet / PDPL / IAS / Cybersecurity Council text may lag the current text — verify against the source.

User Request

$ARGUMENTS

You are an enterprise architect generating a three-tier AI Autonomy Posture for a UAE federal entity AI deployment. This artefact is internal ArcKit synthesis based on the federal three-tier model (Tier 1 internal-productivity, Tier 2 investor-facing-with-approval, Tier 3 regulated/financial); there is no single public regulatory anchor — apply the related UAE AI Charter (AICH) and PDPL obligations as the underlying rules.

Process

1. Read prerequisites:
   • projects/000-global/ARC-000-PRIN-*.md (federal principles, if present)
   • The project's REQ, NFR-SEC, ARCH, and any AICH (UAE AI Charter) artefact (if present)
   • ${CLAUDE_PLUGIN_ROOT}/templates/_partials/RENDERING.md
2. Read the template:
   • First, check .arckit/templates-custom/uae-ai-autonomy-tier-template.md (user override)
   • Then, .arckit/templates/uae-ai-autonomy-tier-template.md
   • Fallback, ${CLAUDE_PLUGIN_ROOT}/templates/uae-ai-autonomy-tier-template.md
3. Use scripts/bash/create-project.sh --json <project-name> if the project does not yet exist; otherwise locate it.
4. Use scripts/bash/generate-document-id.sh AUTI --filename for the artefact filename.
5. Resolve the <!-- DOC-CONTROL-HEADER --> marker per RENDERING.md.
6. Generate the following sections:
   • AI Use-Case Inventory (mapped to tier) — every AI use-case in scope, mapped to Tier 1, Tier 2, or Tier 3 with rationale.
     • Tier 1 — internal productivity (drafting, summarisation, internal Q&A) — low blast radius.
     • Tier 2 — investor-facing or external content, with mandatory human approval before release.
     • Tier 3 — regulated or financial decisions (eligibility, payments, risk scoring) — strict guard-rails, approval gates, audit obligations.
   • Per-Tier Guard-Rail Matrix — for each tier, state the technical guard-rails (input filtering, output filtering, retrieval grounding, hallucination controls, refusal policy, prompt-injection defences).
   • Approval Gates per Tier — who approves what, before deployment and before each release; for Tier 3, the regulator-facing approval where applicable.
   • Audit Obligations per Tier — logging detail (prompt, response, model version, retrieval context, human reviewer), retention, audit cadence, and reporting line.
   • Tier-Promotion Criteria — the criteria that must be evidenced before a use-case is promoted from Tier 1 to Tier 2, or Tier 2 to Tier 3.
7. Populate the External References section per ${CLAUDE_PLUGIN_ROOT}/references/citation-instructions.md. Where the tier framing references the UAE AI Charter, cite the Charter directly.
8. Write the artefact via the Write tool to projects/<project-id>/<filename>.
9. Show only a summary to the user (one paragraph plus the count of use-cases per tier and any tier-promotion proposals).

Authoritative anchor

This artefact is internal ArcKit synthesis lifted from federal practice (no single public regulatory anchor for the three-tier model). The supporting public anchors are:

• UAE Charter for the Development and Use of Artificial Intelligence: https://uaelegislation.gov.ae/en/policy/details/the-uae-charter-for-the-development-and-use-of-artificial-intelligence
• Federal Decree-Law No. 45 of 2021 on the Protection of Personal Data: https://uaelegislation.gov.ae/en/legislations/1972/download

Important notes

• Tier classification is determined by blast radius, not by model capability. A high-capability model used only for internal draft summaries is still Tier 1.
• A use-case may NOT skip tiers — Tier 3 promotion requires evidenced Tier 2 operation.
• Tier 3 use-cases involving financial or regulated decisions may additionally fall under the Central Bank of the UAE supervisory perimeter — flag for sector-regulator engagement and treat that engagement as not-verified within ArcKit until separately confirmed.