This is the public record behind the artifact: what changed, what another person can reuse, what supports the system, and what the build is not entitled to claim.
Started as a form that assumed governance vocabulary.
Became a quick/deep diagnostic with broad-comprehension copy.
Moved into the Exhibition system and added interactive B-Web.
Certification added exact provenance, dimension-level source basis, adversarial fixtures, privacy/support records and a first-class LinkedIn film.
Automated release certification passed 60/60 Playwright checks across Chromium and WebKit on 2026-08-15 (GitHub Actions run 31912745930).
Simplify the interface without simplifying the hidden system.
Treat human presence, timing, evidence, authority, escalation and records as separate control questions.
Classify source-derived rules separately from product heuristics.
NIST AI RMF 1.0 — mapped sections in the Evidence page/manifest.
Regulation (EU) 2024/1689 Arts. 13–14.
UK ICO Artificial intelligence audit toolkit — Human review.
Release gate also passed lineage validation, typecheck, lint and production build before the 60-test browser matrix.
Composite score/thresholds are product heuristics, not compliance rules.
A designed control can still fail in practice.
Applicability depends on jurisdiction, sector and use case.
Automated browser certification is not a substitute for any separately required physical-device or manual review.
VISUAL ASSET REVIEW
The creator-owned cover remains live unless every external-candidate gate is explicitly complete. Disabled controls report canonical status; they do not imply an approval action occurred here.
Status: Promoted after direct-fit review
Official NIST AI Risk Management Framework 1.0, shown as bounded source context.
Credit: Tabassi, E. (2023), Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1. Republished courtesy of NIST.
SHA-256: 88f2b8b92c9c65b0fd0ce66c324bf4936652022bd3193bb70cd3928d2e94b50a
Rights: NIST public information · no endorsement
Supports recognition of the named NIST publication as bounded source context only; does not prove RN prototype performance, compliance, safety, effectiveness, or institutional endorsement.
National Institute of Standards and Technology does not endorse RN or this build.
Disposition: CLEARED-NOT-SUBSTITUTED
CLEARED — exact document asset may be reused under the recorded public-domain/open-license terms, subject to the stated attribution, exclusion and no-endorsement conditions.
Orientation/dimensions: portrait, US Letter; 612 × 792. PDF page box in points; verify page 1 before production crop.
Crop: Crop boundary: PDF page 1: publication identifier, full title, subtitle/version and publication date text; omit all insignia/marks and decorative field. Do not include any interior figure until separately reviewed. Exclude institutional marks, logos, seals, personal imagery and third-party material.
Alt text: Official NIST AI Risk Management Framework 1.0 document or guidance page, selected as contextual evidence for Build 001, Human Review Design Framework.
Credit: Tabassi, E. (2023), Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1. Republished courtesy of NIST.
Claim boundary: Supports recognition of NIST AI Risk Management Framework 1.0 and the limited source context adjacent to Build 001; it does not prove RN's prototype performance, compliance, safety, effectiveness or institutional endorsement.
PROMOTED · LIVE
Direct recognizable-source mapping; readable 1200×1500 production derivative; required attribution, source, claim boundary, rights line and no-endorsement notice remain visible.
Rollback: The prior RN cover remains preserved at its canonical asset URL.
CANONICAL SEVEN-SLIDE PRODUCTION BLUEPRINT
Cover: Open exact mapped asset →
Use the complete live official source-context cover at /media/100-builds/production-source-context/build-001-official-source-context.svg. Do not crop text, credits, safe margins, or claim boundary.
Claim boundary: Supports recognition of the named NIST publication as bounded source context only; does not prove RN prototype performance, compliance, safety, effectiveness, or institutional endorsement.
Accessibility: Slide 1 alt text must identify Build 001, the hook function, essential visible text, and the information conveyed by the visual. Keep reading order headline → visual → boundary/credit.
Rights: NIST public information · no endorsement
RN-owned problem framing: isolate the tension behind “HUMAN IN THE LOOP? That phrase hides the part that matters.”. Use typography plus one simple contrast or decision fork; no stock people.
Claim boundary: RN synthesis only; avoid performance, compliance, safety, effectiveness, causal, testimonial, or endorsement claims.
Accessibility: Slide 2 alt text must identify Build 001, the problem function, essential visible text, and the information conveyed by the visual. Keep reading order headline → visual → boundary/credit.
Rights: Use RN-owned typography, diagrams, and verified prototype captures only.
Show an RN-owned diagram or prototype view of the Human Review Design Framework mechanism. Label inputs, decision/action, and output; do not depict outcomes not present in the build.
Claim boundary: RN synthesis only; avoid performance, compliance, safety, effectiveness, causal, testimonial, or endorsement claims.
Accessibility: Slide 3 alt text must identify Build 001, the system function, essential visible text, and the information conveyed by the visual. Keep reading order headline → visual → boundary/credit.
Rights: Use RN-owned typography, diagrams, and verified prototype captures only.
Use only the recorded official source-context treatment from https://doi.org/10.6028/NIST.AI.100-1; preserve attribution, rights line, no-endorsement language, and exact claim boundary.
Claim boundary: Supports recognition of the named NIST publication as bounded source context only; does not prove RN prototype performance, compliance, safety, effectiveness, or institutional endorsement.
Accessibility: Slide 4 alt text must identify Build 001, the evidence function, essential visible text, and the information conveyed by the visual. Keep reading order headline → visual → boundary/credit.
Rights: Tabassi, E. (2023), Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1. Republished courtesy of NIST. NIST public information · no endorsement
Use an authentic RN prototype/interface capture from Build 001, or a faithful RN-owned diagram if no stable interface view exists. Clearly label simulated, fictional, or illustrative states.
Claim boundary: Show only an observed prototype state; label simulation and fiction; do not imply deployment or outcome.
Accessibility: Slide 5 alt text must identify Build 001, the demo function, essential visible text, and the information conveyed by the visual. Keep reading order headline → visual → boundary/credit.
Rights: Use RN-owned typography, diagrams, and verified prototype captures only.
Use a two-column RN synthesis: what this build makes possible / what it does not establish. No universal, causal, safety, compliance, or effectiveness claim.
Claim boundary: RN synthesis only; avoid performance, compliance, safety, effectiveness, causal, testimonial, or endorsement claims.
Accessibility: Slide 6 alt text must identify Build 001, the implications function, essential visible text, and the information conveyed by the visual. Keep reading order headline → visual → boundary/credit.
Rights: Use RN-owned typography, diagrams, and verified prototype captures only.
Use a quiet RN-owned closing card with Build 001, the canonical dossier URL, CTA, source/credit line, and “RN synthesis · no endorsement” where an institution is named.
Claim boundary: RN synthesis only; avoid performance, compliance, safety, effectiveness, causal, testimonial, or endorsement claims.
Accessibility: Slide 7 alt text must identify Build 001, the cta + sources function, essential visible text, and the information conveyed by the visual. Keep reading order headline → visual → boundary/credit.
Rights: Use RN-owned typography, diagrams, and verified prototype captures only.
Instagram: Publish all seven 4:5 slides; keep credits and boundaries on-slide; write the supplied alt text manually.
LinkedIn: Export as an accessible seven-page PDF; preserve live links in the post copy and use the build dossier as the canonical source.
Reels/TikTok: Animate slides 1–6 as silent-first vertical beats; retain captions, visible source credit, and final CTA; do not imply motion footage is documentary.
Web/newsletter: Use slide 1 as the preview and embed/link the dossier; do not detach the source-context image from its caption, credit, and limitation.
Download all 100 structured blueprints ↓ · Read production instructions →
All seven accessible 1200×1500 slides for this build are generated and included in the deterministic distribution packages. Metadata retains exact alt text, visual guidance, credit, source, rights, and claim boundaries.
Open carousel asset operations → · Open 700-slide inventory → · Read import instructions →
Download this build’s Canva batch ↓ · Download Instagram package ↓ · Download LinkedIn package ↓
This build’s current cover and complete metadata are included in the deterministic 100-cover handoff. The package preserves this dossier’s official/RN decision, source, rights, claim boundary, and rollback state.
Download the complete Canva-ready ZIP ↓ · Open package inventory → · Read import instructions →
A public build record has four sections and they are ordered on purpose. Making is the narrative of what changed, written so that someone who was not there can follow the decisions in sequence. Method is the part meant to travel: the rules and structures another person could apply to a different problem in a different field. Evidence is what supports the system, stated as specifically as the underlying material allows. Limits close the record because they determine how far everything above them can be carried.
The reason limits come last is that they are the part most often left out, and a record without them is a marketing page with citations. Every claim in the sections above is bounded by what the final section says, and a reader who skips it will overstate what the build establishes. If you only have time for two sections, read Method and Limits: together they tell you what you could reuse and how far you could trust it.
Where external sources appear, each is given with an exact locator rather than a bare link, so the specific passage relied on can be checked instead of the whole document. A scope limit accompanies each one. That limit describes what the source was used for in this build, which is frequently narrower than what the source as a whole covers, and quoting the source for a broader claim than the scope allows would be a misuse of it even though the link resolves.
Language here is chosen to hold up under scrutiny rather than to sound confident. Where a review is complete, the record says so and names what the review covered. Where a review remains open, the record says that instead, and continues to say it until the work is done. A statement that something is a demonstration, a synthetic fixture or an unverified claim is a factual description of that artifact, and it stays on the page because removing it would make the surrounding claims read as stronger than the evidence supports.
Finally, this record documents one build among a hundred that are designed to inherit from each other. What Method describes is often the thing a later build depends on, so a record is worth reading alongside the lineage view rather than purely on its own. That is where the capability described here can be traced forward into whatever was built on top of it.
