Storyboards

The book as a sequence of panels: the punchline, the concepts, the evidence and the people they are for, one panel at a time, to flick through on a tablet before a word of a chapter is written. A storyboard is data and the panels are drawn from it by code, so a panel is corrected by changing a line, not by redrawing a picture.

The storyboards

1. The 2018 book, as a storyboard: Book v1.0.0, the 28 published chapters, converted into 15 panels: what the first edition said, who it said it to, and where it stops. (15 panels, draft)
2. A first storyline for the second edition: A proposal in 15 panels: the order the second edition could tell its argument in, built from the topics the newsroom's articles already cover. For Dinis to tear apart. (15 panels, proposal)
3. The opening: you are lucky to be here: The first pages of the second edition as Dinis described them, in 16 panels: wonder, not fear; the history; technology that finally works for you; learning made for you; and why that makes you a developer. (16 panels, draft)
4. Where do you land?: Wardley maps, shown before they are named, and the three centres of gravity: explorer, villager, town planner. 14 panels. (14 panels, draft)
5. Connect the dots backwards: Paths, mixing disciplines, learning as a lifelong skill, and why to read this on paper. 14 panels. (14 panels, draft)
6. Where the thinking came from: The influences behind the book, as claims you can check, and where each one lands. 10 panels. (10 panels, draft)
7. Generative AI, not AI: What the book means by generative AI, how a language model works in plain words, where hallucinations come from, and where Dinis was corrected. 12 panels. (12 panels, draft)
8. Follow the rabbit holes: Provenance, explainability and asking for the evidence; and being the best-prepared person in the room. 10 panels. (10 panels, draft)
9. Everything is connected: What a graph is (not a diagram), the whodunit, a graph becoming a map, groups of groups, provenance, and starting anywhere. 12 panels. (12 panels, draft)
10. Lenses, roles and loops: How LLMs and agents work: a prompt as a lens, a role as a model with a description, an agent as a loop, tools as requests, no memory, compression, and focused threads. 13 panels. (14 panels, draft)

The cast

The cast of the Generation Z Developers storyboards, v0.1, proposed by gen-z.journalist on 18 October 2026 for Dinis to accept, change or replace. Characters are drawn programmatically by assets/storyboard.js from 'look'; nothing here is an image. Each human with a 'gravity' wears its badge (explorer, villager, town-planner), black and white, by their name: a centre of gravity, not a box, as Dinis describes it. Fictional people: each stands for a reader the book is for (the target audience), or for a role in the story. Names chosen to avoid clashing with the riskmandate.ai cast (which has a Maya) and with the word 'developer' (no 'Dev').

How it works

• The data. admin/content/storyboards/cast.json, and one NN__<slug>.json per storyboard: scenes of kind cover, scene, evidence (a quote card or a bar chart), punchline or spectrum; each with a caption, the people in it (who, where, pose, mood, what they say or think), props, a note on what the panel is for, and its source.
• The drawing. assets/storyboard.js draws people, bubbles, captions, props and charts as SVG with Rough.js (MIT, vendored in assets/vendor/), so everything looks sketched; a fixed seed per panel keeps the wobble the same on every load. In the deck the strokes draw themselves and the people and bubbles arrive in order.
• The checks. The build refuses a scene whose character is not in the cast, an unknown pose, mood or kind, or an evidence panel with nothing in it.
• No generated images. Nothing here is drawn by an image model; that keeps every panel correctable, versioned and free of licence questions. Image models can still be given a storyboard as a brief, the way riskmandate.ai's stories do.