Every AINA design era, in one place

~/Downloads/AINA Design Explorations/Luminous OS/ · AI Native Academy — Brand Deck Reference.pdf

The winged mark, gold + teal on warm cream, "Learn · Grow · Command," and the line "Real Learning. Real Practice. Real Impact." The Luminous OS deck (its framing was later ruled ignorable for product) doubles as the cleanest brand-asset showcase.

aina-platform/apps/web/public/mockups/ · live at head.ainativeacademy-web.pages.dev/mockups/ (and, unintentionally, ainative.academy/mockups — see §8)

Twenty mockups across eight screens in three complete styles, with the founder decision recorded on the board itself: "The build is shaped. B primary · A secondary." — B Luminous Dark Glass as app chrome, A Editorial Premium for marketing/dashboard, C Sovereign Bold as campaign wildcard.

Design-lead verdictThe strongest single body of design work in the estate. The B-primary ruling was never fully executed in product — today's app is A-flavored throughout. The three B-mockup ideas (named rail widgets, emerging-shape during assessment, honest "Sample" labeling) are directly liftable into the current product without a re-skin.

aina-platform/apps/web/public/mockups/landing-v2/ · V6 live at head.ainativeacademy-web.pages.dev/mockups/landing-v2/V6-sticky-shell

Six sequential landing concepts, dark-first then light: V1 reskin → V2 chat-as-hero → V3 score-card reveal → V4 product tour → V5 scroll demos → V6 sticky shell (the pi.dev-style sticky product frame that demos the live product as the story scrolls) — the version you ruled canon and that PR #222 ports to the academy.

Design-lead verdictA textbook option ladder — each step answers the previous one's weakness. V6 keeps V2's chat-first conviction, V3's score honesty, V5's light warmth. This history is the argument for finishing PR #222 rather than redesigning again: the exploration was already done, thoroughly.

aina-platform (frozen for merges; still serves the public apex) · https://ainative.academy

The public-site-launch blitz componentized the V6 direction into the live site: Editorial Premium light, "Learn AI. Grow Your Edge. Lead the Future.", six-question assessment, readiness widget. This is what the world sees at the apex domain today, served by the frozen 30-lesson-MVP repo while the academy is built.

Design-lead verdictStill a credible public face — but it sells a six-question assessment and 30-lesson product that the academy supersedes, and its unlabeled sample score is a small honesty debt in production right now. Folding the academy in at the apex is the (founder-gated) end state; until then this page and the academy's V6 landing must not drift apart in message.

aina-academy (this repo) · staging aina-academy-staging.img-playground.workers.dev (access-gated) · tokens in packages/ui + apps/web/src/styles/global.css

The real learner OS: diagnose → path → lesson → practice → progress, Editorial-Premium- flavored light theme on the token set (cream/navy/gold/teal). Fully walked with screenshots and root-caused findings in the polish-lane inventory; its old MarketingLanding still fronts / until PR #222's V6 port lands.

Design-lead verdictThe loop is real and the lesson shell is strong, but the app reads as an A-only product — the B-primary ruling never reached it, and the mockups' honesty details (Sample chips, named greetings, real dimension data) were dropped in translation. The polish lane's Waves 1–3 are exactly the reconciliation of this era with §1's locked decision.

walked 2026-08-06, local build at current main, every off-by-default learner flag forced on (PRACTICE_ARENA_V2, SIMULATION_RUNNER, INTERACTION_VECTORS, FEATURE_STORE, UNIVERSAL_TRACK_COMPOSER…)

The flag registry holds 27 flags; several gate whole surfaces that real users never see. With everything on, the hidden product turns out to be substantial: a complete Practice Arena (briefing → code-editor workspace → deterministic simulation → rubric-scored feedback, with a three-rung hint ladder NUDGE / SCAFFOLD / NEAR-SOLUTION, three modes incl. v2-only Pure Simulation, five control skins — Instructor Studio, Gemini-, ChatGPT-, Codex-, Claude-style — simulation variants, adversarial scenarios, artifact versioning), a Playground of mission cards with success criteria and scored rubric feedback, and a Memory Inspector teaching the L3→L2→L1→source-payload memory chain with an anti-patterns panel.

Design-lead verdictThe hidden 40% of the product is better-designed than several shipped surfaces — the arena's briefing/hints/rubric loop is the most complete pedagogical UI in the estate and already runs scored end-to-end offline. Three asks for the founder queue: an unhide strategy (practice arena is staging_only, playground serves demo missions — what unlocks them for real learners?); the leak pass must include flagged surfaces ("[STUB — Lane D…]", keep_iterating); and /memory's dead chain-load needs a root-cause alongside the P0-4 dead ends. Also noted: the five skins currently differ only subtly — either make them visibly distinct or cut to two.

~/Downloads/AINA Design Exploration 0807/ (+ "(1)"/"(2)" copies in AINA Design Explorations/) · self-contained HTML lab

A live options board over six real surfaces with palette/type/device switchers: five palettes (Warm Cream, Cool Porcelain, Warm Mist, Luminous Ink, Sovereign Bold), three-plus type directions (Cinzel titling enters; Playfair scoped to social), five gold values, CTA treatments. Ships its own SYNTHESIS.md reconciling v1 · v2.1 · the live truth.

Design-lead verdictThe most rigorous token exploration in the estate — this is where the system vocabulary (palette names, gold ramp, type ladder) matured. Superseded as a decision surface by §6's sovereign synthesis, but keep the lab: it's the fastest way to re-test any future palette/type question in real context.

~/Downloads/AINA Design Explorations/0807 … Porcelain Cinzel/ · Ali - Cinzel Design System/aina-design-system-sovereign.html · AINA Cinzel and Porcelain - Design System.html

The v2.1 "Apple-meets-Arc" porcelain direction, its locked-look pass (cream + dark campaign band, teal CTA), and finally the July 16 sovereign edition: "Everything explored since May — three token lineages, five palettes, four type directions — resolved into the register you chose: parchment, warm ink, antique gold, Cinzel titling. Hairlines instead of shadows."

Design-lead verdictSuperseded as "latest word" by the Aug-5 consolidation below — but the July labs are where the vocabulary matured and remain the fastest place to re-test palette/type questions in context.

claude.ai/design project Cinzel and porcelain design system · 10 design canvases + export/ standalone board + aina-tokens-all-themes.css + 9 token-lineage CSS files + brand/UX reference docs · imported read-only 2026-08-06

Ali commissioned Claude Design to consolidate ALL the work — and it did the archaeology with receipts: it recovered Mission Control's real shipped palette from source (finding its DESIGN.md stale), confirmed teal #1AA79D unanimous across Mission Control + V6 + the bound system, found the V6 dark ramp byte-identical in a fourth independent record, and reconciled Porcelain Dark = #07101D + #F5F1E6 from three converging records. Boards cover foundations, marketing site, H1 options, assessment, chat coach, sandbox, score reveal, dashboard, social, lesson-video frames, email, stationery, a permission matrix (five classes incl. a new Console class for founder/ops surfaces), and a data-trust state contract (live / stale / alert / not-wired) adopted as semantic tokens.

Decisions locked in that session: Playfair retired from this direction (Cinzel = titling, Iowan = reading, Inter = chrome, Plex Mono = data) · gold = marketing + one hairline per screen, never a fill · absolutely flat (one exception: the moving gold ring on the single primary CTA) · teal never fills on dark — cream fills, teal means progress/action · waves stay (marketing only), dot grid cut, orbit promoted to the score motif. Its open items: H1 treatment (its pick: Cinzel eyebrow + Iowan headline) · marketing CTA fill teal-vs-navy (its pick: navy on marketing, teal in product) · the gold-gradient divergence vs. shipped V6/Mission-Control CTAs · Mission Control's stale DESIGN.md needs correcting in-repo.

Design-lead verdictThis is now the newest and most rigorous synthesis in the estate — and it is explicitly a deliberate departure from the bound design system (which rejects Cinzel outright). That sharpens the founder question from "Cinzel yes/no" to: adopt Cinzel & Porcelain as the product direction (with its locked rules), or keep the current bound system and scope this to campaign/brand? Everything else in that project (data-trust states, Console permission class, provenance pills, the reconciled dark ramp) is adoptable EITHER way and directly feeds the polish lane — the trust-state contract is practically designed for the honesty fixes in Wave 1.

~/Downloads/AI Native Academy Design System/ (token sheets, ui_kits, surfaces, prototypes) · 30 Lesson Videos Motion Language · AINA Lesson Design System.zip · ui-lesson-previews-ali-recovered-2026-05-18 · aina-mission-control/design · cold-storage/archived-repos-2026-07-12/conductor_repos_aina-design

The packaged v1 design system (color/type/component preview sheets, mobile surface frames, homepage canvas), plus the estates not rendered here: lesson-video motion language, the recovered May lesson-preview UI, internal-tool design languages (mission-control, agent-company), and two archived design repos in cold storage.

Design-lead verdictReference tier. Nothing here should drive new decisions, but the token sheets are the provenance chain for the current UI package — worth keeping exactly where they are, catalogued by this hub, not resurrected.

Concrete, already-designed pieces the current product can lift, mapped to the polish plan's waves:

The one structural observationEvery era converged on the same DNA — warm cream, deep navy ink, gold as achievement, teal as interaction, serif display over sober body. There is no competing design language to choose between; there is one language at different maturities, plus one unexecuted ruling (B-primary app chrome) and one open question (Cinzel). That makes the polish lane's job reconciliation, not reinvention.