HASZB_AIHASZB_AI

Search across courses, lessons, glossary terms, prompts and tools.

Design system

HASZB_AI

Two columns joined by a shared crossbar, with a warm core where they meet — the machine, and the person in the loop, given real dimension.

Identity

One glyph, six expressions. An extruded H: a teal body for the machine, a warm core for the person in the loop.

Symbol

HASZB_AI

Wordmark

HASZB_AIHASZB_AI

Horizontal

Compact / app icon

Monochrome

Favicon size (16px)

Loading

Loading mark

Colour

Warm is human. Cool is machine. Everything else is neutral or status.

Background

Elevated

Surface

Surface 2

Border

Border strong

Primary · machine

Primary soft

Accent · human

Accent soft

Success

Warning

Error

Info

Text

Text muted

Typography

Space Grotesk for display, Inter for reading, JetBrains Mono for code.

Master AI

Build with agents

Understanding context windows

Tokens are not words

A model predicts the next token

Body large — used for lesson introductions and section leads, where the reader needs a slightly larger measure to settle into longer text.

Body — the default reading size across lessons. Readability outranks every aesthetic consideration here: this is a platform people read for hours.

Small — captions, metadata, table cells.

Caption — timestamps, sources, verified dates.

Label · eyebrow · section marker

const response = await model.generate(prompt);

Buttons

Five variants, three sizes, one press response.

Link styled as button

Badges

Status, taxonomy, and the verification stamp used on all dynamic content.

NeutralStage 1Human inputCompletedSimulationNot verifiedVerified 15 Aug 2026

Cards & progress

Stage 2

Understanding LLMs

40

Transformers, tokens, context windows and why models hallucinate.

Course progress40%

Loading

Skeleton state

Forms

Used only for your account.

Describe the task, the context and the output format.

Navigation & overlays

Tooltips appear on focus too

Arrow keys move between tabs; Home and End jump to the ends.

Code

agent.ts
const agent = createAgent({
  goal: "Answer questions using the docs",
  tools: [search, fetchPage],
  requireApproval: ["sendEmail"],
});

Empty & error states

No saved prompts yet

Save a prompt from the library and it will appear here.

Reset your progress?

This clears every completed lesson on this device.

Built on the native dialog element, so focus trapping, Escape and the top layer come from the platform rather than from JavaScript.