# AI Exploration Lab — Design System

> **Tagline:** *"La IA no era para ti. Hasta hoy."*
> **Org:** Universidad Pontificia Bolivariana — Sede Medellín (UPB Medellín)
> **Version:** v1.0 — 2026 · *Documento Vivo* (living document)

The AI Exploration Lab is a university initiative whose mission is to **spark massive
curiosity about artificial intelligence among university students of *any* discipline** —
proving that AI is an accessible, useful, already-available tool, not a future technology
reserved for engineers. The whole brand exists to break one myth: *"AI is for engineers."*

It is **not** a product company. It is a **laboratory of possibilities** ("El Lab no es un
aula. Es un laboratorio de posibilidades"). Every session ends with something *created*.

---

## Sources

This design system was built entirely from the materials below. No codebase or Figma file
was provided — the brand book is the single source of truth.

| Source | Path | Notes |
|---|---|---|
| Brand Book v1.0 (PDF, 14 pp, Spanish) | `uploads/AI_ExplorationLab_BrandBook_v1.0.pdf` | Manifesto, mission, audience, full visual + verbal identity, design tokens, restrictions |
| Logo — dark bg (primary) | `uploads/Logo_Proyecto CLASE AI.png` → `assets/logo-dark-bg.png` | Vector Syntax Glyph + wordmark on Deep Space |
| Logo — light bg (monochrome) | `uploads/Logo_Proyecto CLASE AI_FONDOBLANCO.png` → `assets/logo-light-bg.png` | Cyan→violet glyph for white backgrounds |

> **There is no live product yet.** The brand book specifies *digital applications*
> (landing page, registration forms, presentation slides, social formats). The UI kits and
> slides in this system are faithful implementations of those written specs — not
> recreations of existing screens.

---

## Index / Manifest

Root files:
- **`index.html`** — repo portal (also serves as GitHub Pages homepage)
- **`README.md`** — this file (context, content + visual foundations, iconography, index)
- **`colors_and_type.css`** — all design tokens (color, type, spacing, radii, shadow) + semantic type styles
- **`SKILL.md`** — Agent-Skill manifest for downloading & reusing this system
- **`LICENSE`** — MIT for source · brand assets © UPB Medellín
- **`.gitignore`** — excludes scraps and the regenerable standalone bundle

Folders:
- **`assets/`** — logos (dark + light), isotipo (with transparent variant), Latent Map texture
- **`preview/`** — design-system specimen cards (rendered in the Design System tab)
- **`ui_kits/landing/`** — landing-page UI kit (hero, "cómo funciona", session card, registration form, footer)
- **`slides/`** — presentation slide templates (title, section, content, big-quote, code/data)
- **`uploads/`** — original brand book PDF (kept for reference)

---

## Use this repo

```bash
git clone <your-fork-url> ai-exploration-lab-design
cd ai-exploration-lab-design
# No build step — open files directly in a browser.
open index.html              # or: python3 -m http.server 8000
```

### Deploy the landing on GitHub Pages
1. Push to a GitHub repo.
2. Repo **Settings → Pages → Source: Deploy from a branch → `main` / `/ (root)`**.
3. After ~1 min the portal is live at `https://<user>.github.io/<repo>/`,
   with the landing kit at `/ui_kits/landing/`.

No build, no Node, no bundler — every page works as a static file (React + Babel
are loaded from CDN at runtime).

### Pulling this into another project / Claude Code
Copy the whole folder. The skill is self-contained: `SKILL.md` tells an agent how to
read the system; `README.md` is the human reference; `colors_and_type.css` and
`assets/` are the drop-in primitives.

---

## CONTENT FUNDAMENTALS

**Language.** Primary copy is **Spanish (Colombia)**, conversational register. Technical
docs (this README, tokens) may be bilingual.

**Voice.** *"Habla como alguien que sabe mucho, pero prefiere hacerte preguntas."* Speaks
like someone who knows a lot but would rather ask you a question. Never condescending,
never empty hype. Energetic but honest — *honestidad técnica* is a core value: "No
vendemos magia."

**Person.** Speaks directly to **"tú" / "ti" / "tuyo"** (second person, informal). The
audience is the protagonist: *"para ti"*, *"ya tenías superpoderes"*. The Lab refers to
itself as "el Lab" / "nosotros" sparingly.

**Casing.** Display titles are frequently **ALL CAPS** for impact (see logo wordmark "AI
EXPLORATION LAB"). Mono labels are uppercase with tracking. Body is sentence case.

**Copy formula for social** (from §05):
`[ GANCHO ] + [ EJEMPLO COTIDIANO ] + [ PREGUNTA ABIERTA ] + [ CTA ]`
> *"¿Sabes cómo sabe Spotify qué canción quieres escuchar después? La misma lógica puede
> aplicarse a diagnosticar enfermedades, diseñar productos o entender a tus clientes. ¿Qué
> harías tú si tuvieras esa herramienta? Inscríbete al próximo Lab."*

**Everyday examples, always.** Anchor AI in things the reader already touches: Spotify's
recommendations, the phone autocorrect, the Instagram feed, Maps, Netflix suggestions.
Never abstract.

**Vocabulary**

| ✅ Use (external comms) | ❌ Avoid (external comms) |
|---|---|
| inteligencia artificial, IA | algoritmo, modelo, pipeline |
| crear, explorar, experimentar | deployment, backend, API |
| herramienta, asistente | machine learning (in headline copy) |
| feed, sugerencia, recomendación | disruptivo, paradigma |
| hoy, ya, ahora | innovación tecnológica (genérico) |
| tuyo / tuya, para ti | futuro, próximamente, potencial |

- **Emoji:** not part of the brand. Do not use emoji as iconography. Use the **syntax
  iconography** instead (see below).
- **The word "hoy" ("today")** is load-bearing — it's the hinge of the tagline and the
  whole positioning. Favor immediacy: *hoy, ya, ahora*.
- **Tone to avoid** (real example from book): *"...transformando los paradigmas del machine
  learning y el deployment de modelos a escala industrial."* — jargon that distances the
  non-tech reader.

**Retired tagline — do NOT use:** *"Understand the syntax..."* (it was tech-oriented; the
new tagline is universal across all majors).

---

## VISUAL FOUNDATIONS

**Overall vibe.** Dark, technical, neon-on-deep-space. Reads like a *terminal / latent
space* — confident and engineered, never sci-fi-kitsch. Think contour maps and syntax
brackets, **never humanoid robots**.

**Color.**
- Built on **Deep Space `#0A0F1E`** — a very dark navy, *never pure black* as a fill.
- **Acid Cyan `#00FFF0`** is the single action color: CTAs, links, focus, interactive
  energy. Used sparingly and always on dark.
- **Electric Violet `#8A2BE2`** is the "depth" accent: subheadings, secondary highlights,
  the second note in the logo gradient.
- Text is **white `#FFFFFF`** for headlines, **`#D0D0D0`** softened white for body,
  **`#888888`** muted for meta.
- Semantic: success `#00FF88`, warning `#FFD700`, error `#FF4444`, info `#00B4D8`.
- **Palette discipline:** cyan + violet are accents, not fills. The composition is mostly
  dark surface with small, glowing accents. No pastels, ever.

**Type.** Three families, three jobs, **never mixed on one line**:
- **Plus Jakarta Sans ExtraBold (800)** — display, titles, impact. Often ALL CAPS, tight
  tracking. Hero size 80px.
- **JetBrains Mono (400/500)** — labels, code, data, version stamps, identifiers, kickers.
  Uppercase + letter-spacing for labels.
- **Inter (400/500)** — body copy and long paragraphs only.

**Backgrounds.** Flat **Deep Space**, optionally layered with the **Latent Map** —
a topographic contour-line pattern representing the *latent space* of AI models. Rules:
10–15% opacity over Deep Space, lines in Acid Cyan or Electric Violet, subtle scale, **only
as a background layer, never foreground**, never above 25% opacity, never on light
backgrounds, never rotated/distorted. **No bright/rainbow gradients.** The only "gradient"
in the brand is the cyan→violet stroke transition inside the logo glyph itself.

**Imagery.** *Semantic and real.* Never generic stock, **never humanoid/cybernetic
robots** (this reinforces the exact myth the Lab fights). Imagery, when used, should be
real people creating real things; cool-toned to sit against Deep Space.

**Glow & shadow.** Elevation is expressed as **cyan glow**, not soft gray drop shadows:
`--shadow-glow: 0 0 24px rgba(0,255,240,0.35)`. Subtle `--shadow-sm/md` carry a faint cyan
tint too. Primary CTAs glow.

**Borders.** Hairline. Subtle white `rgba(255,255,255,0.08)` for structure; **Acid Cyan
border on focus** for inputs. Cards sit on `--surface-2 #0D1530` with a faint border, no
heavy shadow.

**Corner radii.** `sm 4px` (chips, inputs), `md 8px` (buttons, cards, code blocks), `lg
16px` (large panels), `full` (pills/avatars). Moderate rounding — engineered, not bubbly.

**Cards.** Background `#0D1530`, 1px subtle border, `radius-md/lg`, optional faint cyan
glow on hover. No colored left-border-accent cards. No emoji cards.

**Spacing.** 4px base scale (1,2,4,6,8,12,16,24). Generous negative space on dark; let the
Deep Space breathe.

**Motion** (not deeply specified in book; derived to fit the vibe): restrained and
precise. Short fades + small translate-ups (8–12px). Easing `cubic-bezier(0.2,0.7,0.2,1)`.
No bounces, no decorative infinite loops on content. Glow can pulse *very* subtly on a CTA.

**Hover / press states** (derived): hover = brighter glow + slight lift, or cyan text on
hover for links. Press = scale 0.98 + dimmed glow. Focus = Acid Cyan border + `--shadow-sm`.

**Layout rules.** Minimalist. Landing = hero with the tagline, a "cómo funciona" section
using syntax iconography, and **one** clear CTA. Posters/flyers: max 20 words, big cyan
title, white body, Latent Map at 15%, QR bottom-right.

---

## ICONOGRAPHY

The Lab does **not** use a conventional pictographic icon set, and **does not use emoji**.
Its iconography is **Syntax Iconography** — characters borrowed from programming syntax as
visual metaphors. Each has a fixed meaning (§04):

| Symbol | Meaning |
|---|---|
| `[ ]` | Container, structure, block of information |
| `{ }` | Data, variables, configurable content |
| `(*)` | Highlight, important note, point of attention |
| `>`   | Action, direction, progress, user input |
| `///` | Separator, transition, start of a new section |

These are set in **JetBrains Mono**, typically in Acid Cyan or Electric Violet. The brand
itself is written framed in brackets: **`[ AI Exploration Lab ]`**. Use these glyphs where
another brand would use icons — as section markers, list bullets, CTA affixes (`Inscríbete >`),
and kickers (`/// CÓMO FUNCIONA`).

**Logo / Isotipo — the "Vector Syntax Glyph".** Rotated, nested diamonds with directional
arrows, stroked in a cyan→violet gradient. Represents *data in motion, depth of layers, and
direction with purpose*. Variants:
- `assets/logo-dark-bg.png` — primary, on Deep Space.
- `assets/logo-light-bg.png` — monochrome glyph for white backgrounds (wordmark greyed).
- `assets/isotipo.png` — **isotipo only** (glyph), square crop on Deep Space. Use for avatars / favicons / nav marks on dark.
- `assets/isotipo-transparent.png` — isotipo with the background keyed out (soft neon glow preserved). Clean on dark surfaces; on light backgrounds prefer the monochrome logo (a faint halo remains).
- *Pure-white* variant is described in the book but was **not supplied** — see Caveats.

Logo rules: clear space = height of the "A" in the wordmark. Never compress, stretch,
rotate, or add shadows. Never place the standard cyan logo on a light background without the
monochrome adaptation. Minimum 32px tall for the isotipo (favicons/avatars/pins).

**Latent Map texture.** `assets/latent-map.svg` is a faithful recreation of the described
topographic contour pattern (no source file existed). Use at 10–15% opacity as a background
layer only. **Flagged as a recreation — see Caveats.**

---

## Caveats / open questions

See the closing note of this README and the on-screen delivery message.
