◆ Open shelves · all welcome

Curious
about AI?

Fifteen courses. From eight-year-olds to regulated industries. Everything is free, everything is real, nothing's on a corporate calendar.

The catalog

For young readers · ages 8–18
Two courses — one for 8–14, a deeper one for 14–18
For ages 8–14

AI for Kids

№01

Foundations through stories and a career-game. The kind of book a parent reads alongside.

5 modules · BeginnerOpen →
For ages 14–18

AI for School Students

№02

How LLMs actually work, prompting as a real skill, deepfakes, and where the line sits on schoolwork.

5 modules · BeginnerOpen →
Core literacy
Read in any order — each stands alone
For everyone

AI Essentials

№03

Vocabulary, history, honest limits. What an LLM is — and isn't — without the marketing.

6 modules · BeginnerOpen →
For builders

AI Deep Dive

№04

Transformers, embeddings, fine-tuning, evals. The trade-offs that don't fit on a slide.

4 modules · AdvancedOpen →
The everyday skill

Prompting

№05

The one skill that improves everything else you do with AI. Built on before-and-after examples, not abstract advice.

5 modules · BeginnerOpen →
Build · Test · Secure
Read in order — this is the path from idea to something real people use
① Build it

Vibecoding

№06

Turn an idea into a working prototype using AI — with no coding background, and knowing what you have actually got.

5 modules · BeginnerOpen →
② Engineer it

Agent Engineering

№07

The scaffolding that turns a model into a working system — loops, context, tools, and the failures that only appear in production.

5 modules · AdvancedOpen →
③ Test it

Does Your AI Actually Work?

№08

An eval is a test for AI. Build real test cases, judge quality honestly, and catch problems before your users find them.

4 modules · AdvancedOpen →
④ Secure it

Securing AI Systems

№09

Threat-model, harden and red-team AI and agentic systems. The hands-on counterpart to the sales course.

4 modules · AdvancedOpen →
⑤ Run it

From Prototype to Real Product

№10

The honest second half. Data, accounts, shipping, cost, support and the second year — with the effort counted at every stage.

5 modules · BeginnerOpen →
Industry courses
Vertical, role-specific. More on the way.
For security teams

AI for Cybersecurity

№11

AI in the security stack and the security of AI itself — defence, the adversarial side, and how to evaluate the products.

4 modules · IntermediateOpen →
For clinical & health-tech

AI for Healthcare

№12

Where AI actually sits in care, how to read the evidence, and the regulatory reality. Orientation, not clinical guidance.

4 modules · IntermediateOpen →
For legal practice

AI for Legal

№13

Contracts, discovery, and the citation problem that has already cost lawyers their credibility.

4 modules · IntermediateOpen →
For risk, compliance & credit

AI for Finance

№14

AI inside the model risk duties finance already has — explainability, fair lending, and the failure modes that cost money.

4 modules · IntermediateOpen →
For auditors & assurance teams

AI for Audit

№15

AI on both sides of the engagement — using it without surrendering scepticism, and auditing numbers a model helped produce.

4 modules · IntermediateOpen →
In the kiln
More verticals next quarter
Healthcare, legal, and field-sales drafts are in review. Suggest one →
Run a cohort with us
Bring your team
We co-author industry courses with practitioners. If that's you, get in touch.

Side stuff

Two rabbit holes · enter at your own risk
Try things · break things

AI Playgrounds

§

A wall of small, runnable toys. Tokenizer in your browser. Embeddings on a 2D plane. Ten minutes teaches you what an hour of reading wouldn't.

  • Next wordWatch a sentence get predicted, one word at a timeRUN →
  • Base rateWhy an accurate detector still cries wolfRUN →
  • Eval confidenceIs that improvement real, or is it noise?RUN →
  • TokeniserSee how your sentence becomes numbersRUN →
7 toys · all in-browser · no API keysOpen the room →
Field notes · not yet a course

Agentic AI

Tools, memory, the loop. Watch an agent reason, call tools, and act — an interactive anatomy plus three step-by-step investigation scenarios.

  • I.The agent loop — how agents think and actanatomy
  • II.Research: when the sources disagreescenario
  • III.Trip planner: stopped at the payment gatescenario
  • IV.Bug hunt: the first hypothesis was wrongscenario
1 interactive anatomy · 3 scenariosRead the notes →
Marginalia · Did you know

GPT stands for Generative Pre-trained Transformer.

Each word does real work. Generative: it produces, doesn't classify. Pre-trained: the heavy lifting happened before you arrived. Transformer: the 2017 paper that changed everything.

AcronymsTransformer2017
A free public library for AI literacy. Built by people who'd rather you understand than subscribe.
Developed by Salih AA
RSSNEWSLETTERSOURCEEST. MMXXV