# Final editorial QA — evening AI architecture lane

## Deliverable

- **Calendar date:** 2026-09-06
- **Lane:** Evening AI architecture
- **Title:** *The Alias Is Not the Model: Prove Which AI Runtime Touched Facility Work*
- **Author:** Jared Mastroianni
- **Topic:** Request-time runtime identity, mutable model aliases, fallback eligibility, staged authority and post-release monitoring for AI-assisted self-storage work.
- **Proposed slug:** `ai-model-alias-runtime-identity-self-storage`
- **Selected owned destination:** Jared's personal authority site only
- **Publisher owner:** Jared Mastroianni
- **Proposed canonical:** `https://jaredmodstorage.github.io/writing/ai-model-alias-runtime-identity-self-storage/`
- **Non-selected owned publisher:** modSTORAGE WordPress blog; it must reject this full article, title, slug and body as a duplicate.
- **Body word count:** **2,197**. Counting rule: Unicode letter/number tokens with internal apostrophes and hyphens counted as one; bounded by the `BODY START` and `BODY END` markers; includes body headings and lists; excludes Markdown footnote markers, title, byline, reference entries and disclosure.
- **Article SHA-256:** `31247b34d50034de479bca82a2513c367b546c4e575952e5761f654477773540`

## Completeness and operator usefulness

- The deliverable is a complete 2,197-word article inside the requested 1,800-to-2,500-word range, not a topic list, outline, research memo or strategy document.
- The article separates requested, resolved and approved runtime identity; fixes the surrounding operating contract; distinguishes structural from behavioral evaluation; uses shadow traffic before authority; stages release by authority mode; and defines safer fallback and post-release monitoring.
- `ai-runtime-identity-receipt.csv` parses at 45 consistent columns across one blank template row and five explicitly fictional teaching records. Its five nonblank receipt IDs are unique. Tool SHA-256: `d2036b2a66dcb99eef94f8c3cca505f1ca3df0815b79b49f5ae536a77612ba70`.
- Copper Vale Storage and every associated facility, model, provider, router, system, person, record, timestamp, test and outcome are explicitly fictional.

## Source and factual review

- Eight current primary or authoritative source records are included: the current NIST AI RMF page, NIST AI 800-4, the original model-cards research page, current MLflow Model Registry documentation, JSON Schema 2020-12 Core, two current AWS SageMaker deployment pages and OpenTelemetry's GenAI span specification pinned to commit `94f432d7`.
- All eight registered URLs returned HTTP 200 after redirects on 2026-09-06. The source register parses at eight unique source IDs and 11 consistent columns. Source-register SHA-256: `d80d96815d22ce3518c604d5382070478fb17a2ca0e8c66ff14d6c8046baf030`.
- NIST AI RMF 1.0 is accurately described as voluntary and under revision. NIST AI 800-4 is bounded as a cross-sector monitoring landscape whose practices and terminology remain developing.
- MLflow, JSON Schema, AWS and OpenTelemetry support only their documented registry, schema, deployment and telemetry mechanics. OpenTelemetry fields are marked as development-state. None is presented as proof of business equivalence, self-storage suitability, privacy compliance, deployment or operating outcome.
- Article references 1 through 8 map one-to-one to source-register records S1 through S8. No unregistered external factual citation remains.

## Duplicate, originality and voice review

- Preflight covered the current 46-record calendar, 75-record editorial/media register, 358-record submission tracker, 343-record published-asset register, 551-record evidence register, current-day morning and midday packages, other-lane memories, prior evening packages and active or submitted premises.
- Human-review queue capacity was rejected because it collides with a published canonical and active external routes. Broad fallback/substitution framing was narrowed after external screening found closely related generic technical language.
- Live public readback covered 58 Jared authority-site canonicals and 139 modSTORAGE post canonicals. Neither sitemap contains the selected slug; both proposed URLs returned HTTP 404 and the WordPress exact-slug query returned `[]`.
- Originality screening covered 290 unique-content relevant Markdown files: zero normalized 14-word sentence matches; maximum six-word-shingle overlap 0.179% (four generic NIST or fictional-example phrases).
- Generic-language scanning found none of the prohibited filler phrases listed in the originality review. The article uses no fake quotation, inflated product claim or self-aware writing commentary.

## Destination and state review

- `destination-assignment.md` was created before `owned-publication-manifest.md` and assigns exactly one owned publisher, proposed slug, canonical and publisher owner.
- No WordPress publisher manifest was created. The non-selected modSTORAGE publisher must reject the complete title, slug and body as a duplicate.
- The preflight canonical was absent. Current governance authorizes one reversible personal-authority-site release after the exact source, rendered, accessibility, deployment and public-readback gates pass. Draft, repository change, deployment, live publication, sitemap/feed inclusion, crawling, indexing, ranking, coverage and recognition remain separate states.
- No cold outreach, payment, external submission, CMS mutation, repository deployment or irreversible external action occurred.

## Three editorial review passes

These are distinct internal editorial passes by the publishing workflow, not independent external review, outlet acceptance or recognition.

1. **Argument and operator utility — 9.7/10.** The article presents a complete operational argument, distinguishes requested, resolved and approved runtime identity, and gives operators an actionable 45-field receipt, release sequence and fallback rule without pretending that provider metadata proves behavior.
2. **Evidence and claim containment — 9.8/10.** All eight factual references map one-to-one to current primary or authoritative sources. The article preserves the voluntary, developing, provider-reported, fictional and non-deployment boundaries and makes no unsupported customer, performance, product or certification claim.
3. **Structure and publication craft — 9.6/10.** The headline, opening, section order, tables, fictional case, monitoring section and closing rule form a coherent expert paper. The prose remains readable for operators while retaining the technical precision needed by architects and reviewers.

4. **Rendered visual review — 9.7/10.** The exact integrated candidate was reviewed at 1,440, 390 and 320 pixels. The H1, lede, byline, governed image and caption, companion-download module, body hierarchy, sources, disclosure, related reading and footer remain contained and legible. Document width equals viewport at all three sizes, the 3,840 × 2,560 image resolves with the exact alt and caption, and the review found no clipping, overlap, missing content, console error or page error.

## Image and file review

- Governed image 052 was visually reviewed at 3,840 × 2,560 and copied byte-for-byte from its high-resolution source. Source and package SHA-256 values are both `5aa6a224294c7426eba866e92c569e6216ff2ea0d05820e1426977ebebe293e1`.
- The image is disclosed as Jared-owned AI-generated editorial material, not documentary evidence of a team, facility, model, provider, system, deployment or result. Its August 27 and September 1 package use is disclosed.
- No author portrait was used. The approved navy-blazer v2 portrait was not needed. The rejected black-turtleneck portrait and black-turtleneck imagery were not used.

## Final editorial note

The package is approved for one reversible release through the proven Jared authority-site source/deployment pair once the remaining rendered, deployment and public-readback gates pass. House style, visible image and assistance disclosures, protected-evidence and retention rules, provider-reported metadata limits, and the release policy that maps model-task evidence to permitted authority must remain visible. The CSV is an adaptable operating control, not a certification, provider contract, privacy assessment, security standard or authorization to process sensitive data.

**Truthful state:** Owned publication is authorized and source integration is in progress. No repository commit, deployment, live publication, canonical activation, search submission, crawling, indexing, ranking, coverage or recognition is established by this editorial checkpoint.