← Library

Arcade β€” stage index

Dev test surface: 31 lessons with a arcade stage β€” each link opens the rendered player.

Building AI Fluency

Data Portraits

Who Do They Think You Are?

Encounter the four criminal charges against humanity before any teaching begins β€” create cognitive dissonance, not comfort

Building AI Fluency

Hello World

The Last Human Job

Judge which jobs survive automation

Building AI Fluency

Questioncraft

Watch the AI Break

Students rewrite a soft, closed interview question into a sharp, open, undodgeable one against a commit-clock, then the room votes the sharpest line of each round, proving that a closed question hands the subject an exit while a sharp one pins them to a real answer.

Building AI Fluency

Writers Room

Bridge of Fake News

Students snap-judge ten real AI press headlines as REAL or HYPE under a deadline clock, then confront a scored credulity meter that exposes their own bias β€” buying staged demos as real while dismissing the quiet peer-reviewed breakthroughs as hype.

Building AI Fluency

Six Lenses

The Trolley

In 5 minutes, get the room to agree to the rules of hard ethical thinking before the dilemma hits β€” so the debate that follows is honest, not point-scoring.

Building AI Fluency

Steelman

The Docket

Students seat themselves as the bench on the real Garcia v. Character Technologies wrongful-death docket, read the unsoftened facts of record, and commit a locked liability verdict on a single legal doctrine before the real court disposition is revealed against their stance.

Building AI Fluency

Teachers Conference

Memory Heist

Students memorise a wall of converging words under a clock; the vault seals; they snap-judge a lineup IN/OUT β€” and catch their own memory stamping a word it never saw as "seen", exposing reconstructive memory (the DRM effect) before any teaching.

Building AI Fluency

The Revision Sprint

Brain Teaser

Students retrieve the week's core AI concepts from memory under a tight clock, then read a per-topic heat-map that exposes which topics they only recognised rather than actually recall.

Creating a Trailer

The Concept β€” Brief, Examples & Idea Lock

Film Ideation

Wake the story instinct: every story is someone who wants something, with something in the way.

Creating a Trailer

The Trailer Logline

The Trailer Logline

Students draw three random genre cards into Snyder's protagonist/goal/stakes slots and assemble a one-sentence logline under a 30-second timer, then the class swipe-votes it watchable or skip β€” producing a messy pitch first so the three-slot logline spine can be formalised after (Kapur productive failure).

Creating a Trailer

Trailer Beats β€” Grammar of the Cut

Trailer Beats

Students reorder seven shards of the original Matrix (1999) teaser into the order a trailer house cut them, then score their cut against the real one to discover that a teaser is an engineered escalation of beats, not a highlight reel.

Creating a Trailer

Who Does AI Work β€” & What Does It Cost?

Pick Your Tools

Guess real API costs from usage descriptions

Creating a Trailer

The Style Frame

Moodboard & Style Frame

Students read six real film stills on instinct, match each to the mood it carries, and argue one contested frame β€” discovering that mood is built from nameable composition (thirds, leading lines, symmetry, light) rather than magic.

Creating a Trailer

Consistent World & Character

Consistent Characters

Train the eye to catch a single continuity break across a six-panel sweep, then reveal the four axes caught β€” palette, costume, world-prop, character-face β€” as the exact fields the next stage makes students author to keep a world and a lead consistent.

Creating a Trailer

Hero Shots β€” Shotlist & Keyframes

Hero Shots & Cinematography

Grasp why a story rides on a few frames, and how to choose them.

Creating a Trailer

Image β†’ Motion

Image β†’ Motion

See how fast AI video improved β€” and what still breaks.

Creating a Trailer

Edit β€” Cut to the Beat

The Edit

Students slam a marker on a sweeping playhead each time they feel a shot should cut on the beat, then score their frame-accuracy against the editor's real cuts β€” feeling, before any theory, that a cut is a timed decision against the music.

Creating a Trailer

The Voice β€” Narration & VO

The Voice

Hear what a voice does to a cut β€” and face the rights fight over cloning it.

Creating a Trailer

The Score β€” Music & Sound Design

Score & Sound

Feel how score and sound carry emotion the image alone cannot.

Creating a Trailer

Finish & Screen

The Film Festival

Under a sub-second lock clock, students read the engineered mood of animated title cards from typography and motion alone β€” kerning, slide direction and pacing β€” scoring on speed and accuracy before discovering the feeling was authored, not read.

Coding JARVIS

The Hacker Toolkit β€” Set Up the Dev Loop

The Design Sprint

Students predict pass-or-crash for AI-written Python, run it at a faux terminal, read the real traceback, patch the one broken line and re-run it green — feeling the predict→run→read→patch→re-run developer loop before it is taught.

Coding JARVIS

Vibe-Coding a First Feature

Vibecode Hackathon

Grasp what "vibe coding" really means β€” and where it bites back.

Coding JARVIS

Persona Design

Persona Design

Students match three anonymised AI outputs to one fixed prompt back to the three system-prompt personas that produced them, discovering that a model's voice, hedging and refusal behaviour are set by its system message β€” same engine, different mask.

Coding JARVIS

Building a Knowledge Base

Building a Knowledge Base

See why ungrounded models invent facts β€” and what grounding actually buys you.

Coding JARVIS

Agent Tool Use β€” Give It Hands

Agent Tool Use

Understand why a model alone cannot act, and how tools give it hands.

Coding JARVIS

Multi-step Agent Loops

Multi-step Agent Loops

Students parse an authentic ReAct agent transcript line by line β€” tagging each line Thought, Action, Observation, or Final Answer β€” so they can see, before any instruction, that an agent answers by looping (reason, act, observe, retry) rather than producing one clever response.

Coding JARVIS

Agent Ethics β€” When Should JARVIS Refuse?

Agent Ethics

Make a moral call under the clock before you reason about who is liable.

Coding JARVIS

Adversarial Testing β€” Break Your Agent

Adversarial Testing

Classify real, attributed jailbreak prompts into the OWASP LLM Top-10 2025 attack buckets against a per-attack countdown, scoring speed times accuracy on a live leaderboard, so students activate the schema they already hold before being taught how the attacks work.

Coding JARVIS

Deploy the Chat Widget

Deployment

Order a real Hugging Face Spaces deploy pipeline under a launch clock, surfacing the universal misorder (pushing before the secret is set) and the mines that look like steps, so the room feels why a build ships or fails before the deploy workflow is taught.

Coding JARVIS

Demo at Seoul Expo

Final Presentations

Students judge three live agent demos as falsifiable claims β€” ranking, scoring each on a visible-proof grid, and betting which founder got funded β€” then confront that polish without proof loses, before any teaching.

Coding JARVIS

Your Future CV

Your Future CV

Run a 4-round interview quality game β€” students evaluate answer strength, the flip reveals the verdict.