AI Compendium · Figure Catalog (art SSOT)
Single source of truth for the illustration phase. Lists the 61
F-*figures of the Compendium, classified by production method and status. Just as the 8-section template organized the text (Part IV), this catalog organizes the art.
Art direction
- Palette:
#1d3557(deep blue) ·#457b9d(medium blue) ·#a8dadc(lightblue) ·
#e63946(accentloopalert) ·#f1faee(background). - Figure typography: sans-serif (Inter/Segoe UI); title 22 pt bold,
captions 11–13 pt.
- Formats: SVG (web, vector, source of truth) → PNG/falling into the PDF via
the D2/WeasyPrint toolchain. Photos in high resolution (≥1500 px wide).
- Assets: SVGPNG live in `assets` within each part (e.g.,
01-ciclo-de-vida/assets/F-I.1-ciclo-de-vida.svg). D2 diagrams stay inline and render directly (they don't appear here). - Metaphor mark: whenever a figure represents a metaphorical concept, indicate it
visually (dashed / translucent outline), as in
F-I.5.
Production methods
| Method | Meaning | Who produces |
|---|---|---|
| svg | Diagram/infographic authorable by hand in SVG | this session / AI |
| illus | Pictorial/artistic scene | image generator or illustrator |
| photo | Photo of a real historical person/object | find public domain / Wikimedia (check license) |
| d2 | Already an inline D2 diagram (renders) | toolchain — no asset needed |
Current status — 31 SVGs ready and render-verified
- ✅ Part I (5/5): F-I.1 cycle · F-I.2 conveyor · F-I.3 icons · F-I.4 heat
map · F-I.5 three layers.
- ✅ Part II (1/1): F-II.1 mural (6 paradigms × cards).
- ✅ Part III (frieze): F-III.0 frieze of 7 eras. (F-III.1–19 = photos/scenes, external.)
- ✅ Part IV (35/35): 24 flow anatomies (common generator) + 11
special hand-authored (Transformer 1.0, LLM cycle 1.1, GAN 5, GNN 11, multimodal 14, knowledge graph 17, Bayesian DAG 19, GP curve 21, forest+SVM 22, swarm 25, MCTS tree 27). All render-verified.
- ⬚ illus (~7) and photos (~12) from Part III — the only outstanding items, and they are
external steps (image generator / Wikimedia curation).
42 SVG figures from Parts I–IV/VII ready. The Part III archive remains (photos/scenes —
art-briefs.kmd) and the figures of the new Parts V and VI (below, still as briefs).
Parts V–VI and state of the art ✅
Added during the integration of the former Volume (2026-06-27); 14 SVGs authored and render-verified (constellation, workshop, levers, funnel, pyramid, workbench, panel, gears, flywheel, knowledge-hub, proof, material, life, product):
- Part III:
F-III.20constellation of 2026 models (svg/illus). - Part V — Building AI:
F-V.0workshop ·F-V.1levers at scale ·F-V.2data funnel ·F-V.3pyramid of builds ·F-V.4toolbench ·F-V.5evaluation panel ·F-V.6why code wins ·F-V.7Kode flywheel. - Part VI — Domains:
F-VI.0AI crossing disciplines ·F-VI.1verifiedproof ·
F-VI.2equation→material ·F-VI.3life folded/generated ·F-VI.4foundation→local product.
Part I — Life Cycle
| Figure | Doc | Method | Status |
|---|---|---|---|
| F-I.1 | 01-ciclo-de-vida/INDEX | svg | ✅ assets/F-I.1-ciclo-de-vida.svg |
| F-I.2 | 01/01-visao-geral | svg (artifact conveyor) | ✅ assets/F-I.2-esteira-artefatos.svg |
| F-I.3 | 01/02-etapas | svg (sheet of 11 icons) | ✅ assets/F-I.3-icones-etapas.svg |
| F-I.4 | 01/03-ciencias | svg (triptych heat map) | ✅ assets/F-I.4-mapa-de-calor.svg |
| F-I.5 | 01/04-natureza | svg | ✅ assets/F-I.5-tres-camadas.svg |
Part II — Types of AI
| Figure | Doc | Method | Status |
|---|---|---|---|
| F-II.1 | 02-tipos-de-ia/INDEX | svg/illus (card mural) | ⬚ |
Part III — Timeline
| Figure | Doc | Method | Status |
|---|---|---|---|
| F-III.0 | INDEX | svg (historical frieze) | ⬚ |
| F-III.1 | 01-precursores | photo — Alan Turing | ⬚ |
| F-III.2 | 01-precursores | photo — Claude Shannon | ⬚ |
| F-III.3 | 01-precursores | illus (from logic to machine) | ⬚ |
| F-III.4 | 02-nascimento | photo — John McCarthy | ⬚ |
| F-III.5 | 02-nascimento | photo — Perceptron Mark I / Rosenblatt | ⬚ |
| F-III.6 | 02-nascimento | illus (Dartmouth summer) | ⬚ |
| F-III.7 | 03-invernos | photo — Geoffrey Hinton | ⬚ |
| F-III.8 | 03-invernos | photo — Yann LeCun | ⬚ |
| F-III.9 | 03-invernos | illus (ice and rules) | ⬚ |
| F-III.10 | 04-aprendizado | photo — Vladimir Vapnik | ⬚ |
| F-III.11 | 04-aprendizado | photo — Fei-Fei Li | ⬚ |
| F-III.12 | 04-aprendizado | illus (the tide of data) | ⬚ |
| F-III.13 | 05-revolucao | photo — Ilya Sutskever | ⬚ |
| F-III.14 | 05-revolucao | photo — Demis Hassabis | ⬚ |
| F-III.15 | 05-revolucao | illus (the cat that changed everything) | ⬚ |
| F-III.16 | 06-transformers | photo/illus — Transformer authors / attention | ⬚ |
| F-III.17 | 06-transformers | svg/illus (attention mechanism) | ⬚ |
| F-III.18 | 07-multimodal | photo/illus — Anthropic / logos | ⬚ |
| F-III.19 | 07-multimodal | illus (converses, creates, and acts) | ⬚ |
Part IV — Chapters (one anatomy figure per chapter)
All F-IV.<cap>.0 (and the extra F-IV.1.1 for the LLM) are schematic svg (type anatomy/flow diagrams). 35 figures, all ⬚ pending:
F-IV.1.0 Transformer anatomy · F-IV.1.1 LLM cycle · F-IV.2.0 noise→image · F-IV.3.0 feature hierarchy · F-IV.4.0 patches · F-IV.5.0 G-D duel · F-IV.6.0 latent funnel · F-IV.7.0 wave→text · F-IV.8.0 text→voice · F-IV.9.0 prompt→song · F-IV.10.0 temporal coherence · F-IV.11.0 messages in a graph · F-IV.12.0 temporal loop · F-IV.13.0 state sweep · F-IV.14.0 one brain, many senses · F-IV.15.0 rules · F-IV.16.0 A* search · F-IV.17.0 entities-relations · F-IV.18.0 formula→solution · F-IV.19.0 belief that propagates · F-IV.20.0 hidden states · F-IV.21.0 prediction with uncertainty · F-IV.22.0 forest+SVM · F-IV.23.0 selection across generations · F-IV.24.0 networks that evolve · F-IV.25.0 swarm · F-IV.26.0 learning by reward · F-IV.27.0 MCTS · F-IV.28.0 actor-critic · F-IV.29.0 feedback→ behavior · F-IV.30.0 perceive-think-act · F-IV.31.0 search→answer · F-IV.32.0 intuition+proof · F-IV.33.0 two-tower · F-IV.34.0 perception→action.
Production summary
| Method | Qty | Note |
|---|---|---|
| svg | ~42 | Authorable by the AI; 2 ready (F-I.1, F-I.5) |
| illus | ~7 | Pictorial scenes — require an image generator/illustrator (not available in this session as a tool) |
| photo | ~12 | Find public domain / Wikimedia; verify license before publishing |
Pipeline honesty: schematic SVGs the AI produces here. The illus scenes and the photos require, respectively, an image generator/illustrator and curation of free archives — steps external to this session. The catalog leaves each figure ready to commission.