How this course works
The course moves through six levels that mirror the UNESCO progression of Understand → Apply → Create. You don't jump ahead: each level ends with a milestone project, and you must reach Proficient (level 3 on the rubric) before starting the next level. That single rule is what turns "watched some videos" into "can actually do the work."
The weekly rhythm — repeat this every week
Roughly 8–10 hours, split across the week so it never feels heavy:
| Phase | Sessions | What you do |
|---|---|---|
| Learn | 2–3 | Work through that week's lesson / video / reading. Take notes in your own words. |
| Build | 2 | Do the hands-on task. You learn AI by making things, not by watching. |
| Log | 1 | Post what you built to your public Build Log. Write 3 lines: what, how, what broke. |
A note for the mentor (the older brother)
Your job is to set the week's task and then get out of the way. Don't do the work for him — review his Build Log, ask "why did you do it that way?", and grade his milestone honestly against the rubric. The fastest way to ruin this is to rescue him every time he's stuck. Let him struggle for 20 minutes first; that struggle is the learning. Where you help most: give him a real task on one of your live projects once he reaches Level 2.
LEVEL 0 · WEEKS 1–2Orientation & Setup
By the end you can
- Explain in plain language what AI, machine learning and an LLM are — and what they cannot do.
- Describe the main ethical issues: bias, privacy, misinformation, and the effect on jobs.
- Operate a professional toolkit: a clean email, GitHub, a live portfolio, and a public Build Log.
Weekly plan
| Week | Focus |
|---|---|
| Week 1 | What is AI / ML / LLMs. The AI landscape by category (text, image, video, voice, code, automation). Capability vs. hype. |
| Week 2 | Ethics & societal impact — bias, privacy, deepfakes, jobs. How to use AI responsibly and honestly. |
Free resources
- Elements of AI — Ch. 1–2 FREE
- Anthropic AI Fluency — Intro FREE
- Google AI Essentials — Module 1 FREE trial
Assignments
AI Landscape Map
Make a one-page visual that sorts 15 real AI tools into their categories, with one line each on what it's for.
Ethics Reflection
Write 500 words: one concrete benefit and one real risk of AI, each with a real-world example you researched.
Your professional home base is live
A working portfolio page (GitHub Pages or Carrd) with a short intro, plus a public Build Log containing your first two artifacts. This is the foundation everything else gets posted to.
LEVEL 1 · WEEKS 3–8AI Fluency & Prompt Engineering
By the end you can
- Apply the 4D framework — delegate the right task, describe it clearly, judge the output, and use it responsibly.
- Write structured, reliable prompts using roles, context, constraints, examples and forced output formats.
- Use AI to research, write, analyse and learn far faster — while catching its mistakes.
Weekly plan
| Week | Focus |
|---|---|
| Week 3 | Delegation & Description — choosing what to hand to AI, and framing the task so it succeeds. |
| Week 4 | Prompt patterns — role prompting, few-shot examples, step-by-step reasoning, structured output. |
| Week 5 | Discernment — evaluating answers, spotting hallucinations, fact-checking, knowing when AI is wrong. |
| Week 6 | Diligence — privacy, honesty, citing AI use, avoiding plagiarism and over-trust. |
| Week 7 | AI for real work — research reports, summarising, extracting data, drafting and editing. |
| Week 8 | Project week — build and polish the milestone toolkit. |
Free resources
- Anthropic AI Fluency — full course FREE
- Elements of AI — Ch. 3–4 FREE
- Anthropic prompt engineering guide FREE
Assignments
Personal Prompt Library
Build and document 20 reusable prompts you'd actually use (study, writing, research, coding help). Each with a note on when to use it.
Prompt Challenge Set
Solve 10 given tasks. For each, show your first attempt, what went wrong, and the improved prompt. The improvement is the point.
AI Productivity Toolkit for a real person
Pick a persona (a shopkeeper, a student, your mother) and build a documented set of AI workflows that saves them time. Include a short write-up: who it's for, what it does, and proof it works. Graded against the rubric — reach Proficient to advance.
LEVEL 2 · WEEKS 9–14No-Code AI Building
By the end you can
- Build a working chatbot / assistant with a knowledge base — no code.
- Automate a multi-step real-world workflow that connects AI to email, sheets, or a website.
- Understand APIs, JSON and webhooks well enough to wire tools together confidently.
Weekly plan
| Week | Focus |
|---|---|
| Week 9 | How AI apps connect — APIs, JSON, webhooks explained simply (concepts, still no code). |
| Week 10 | Build a chatbot with a knowledge base using a no-code builder. |
| Week 11 | Intro to automation with n8n / Make — triggers, actions, AI nodes. |
| Week 12 | Automation #1 — auto-summarise & sort incoming emails or articles. |
| Week 13 | Automation #2 — a small content or data pipeline end-to-end. |
| Week 14 | Project week — build the real client case study. |
Free resources
- n8n docs + YouTube course FREE
- Make.com tutorials FREE tier
- Chatbot builder docs FREE tier
Assignments
Published Chatbot
A chatbot for one specific use-case (e.g. a shop's FAQ), shareable via a link.
Automation + Time-Saved Analysis
One working automation, plus a short "before vs. after" showing how much time it saves and for whom.
Automate a real task for a real business
Find a real or local business (even a family shop) and automate one genuine task for them. Write it up as a proper case study: problem → solution → result. This is his first portfolio proof that he can deliver.
LEVEL 3 · WEEKS 15–22Programming Foundations for AI
By the end you can
- Write basic Python — variables, logic, loops, functions, and reading/writing files.
- Call an LLM API (Claude / OpenAI) from your own code and handle its response.
- Build and ship a small AI-powered script or app with a clean README.
Weekly plan
| Week | Focus |
|---|---|
| Week 15 | Python I — setup, variables, types, input/output. |
| Week 16 | Python II — conditionals and loops. |
| Week 17 | Python III — functions, lists & dictionaries, files. |
| Week 18 | APIs & JSON in Python with the requests library. |
| Week 19 | Your first LLM API call — messages, roles, parameters. |
| Week 20 | Structured output, error handling, tokens & cost. |
| Week 21 | Build a small AI app (a simple Streamlit web UI). |
| Week 22 | Project week. |
Free resources
- freeCodeCamp — Python for Beginners FREE
- Harvard CS50 AI — selected weeks FREE
- Anthropic Claude API docs FREE
Assignments
Python Mini-Exercises
Complete 5 small programming exercises covering logic, loops, functions and files.
Your First AI Script
A Python script that calls the LLM API to do something genuinely useful (e.g. summarise a folder of text files).
A small AI app on GitHub
Build one AI-powered app that solves a real problem, with clean code and a README that explains it. This proves he can build, not just automate.
LEVEL 4 · WEEKS 23–28Real AI Systems — RAG & Agents
By the end you can
- Explain and build a RAG system that answers questions from your own documents.
- Use a vector database and improve retrieval quality.
- Build a simple agent that makes decisions and uses tools — and evaluate whether it actually works.
Weekly plan
| Week | Focus |
|---|---|
| Week 23 | What RAG is and why businesses pay for it. Embeddings, explained simply. |
| Week 24 | Build a document Q&A — load, chunk, embed, retrieve, answer. |
| Week 25 | Vector databases (e.g. Chroma) + improving retrieval quality. |
| Week 26 | Agents & tool use — letting AI take actions safely. |
| Week 27 | Evaluation — test sets, measuring quality, cutting cost. The most underrated skill. |
| Week 28 | Project week. |
Free resources
- LangChain / LlamaIndex intro FREE
- RAG tutorials (freeCodeCamp) FREE
- Chroma vector DB docs FREE
Assignments
Working RAG Q&A
A document-Q&A system over a real set of documents (a manual, a set of notes, a policy PDF).
Evaluation Report
Write a test set of questions, run it against your RAG, and report where it's right, wrong, and why.
A "Knowledge Assistant" over real business documents
Build a RAG assistant on a real organisation's documents, with an evaluation showing it actually answers correctly. This is a portfolio piece that gets people hired.
LEVEL 5 · WEEKS 29–32Becoming a Consultant
By the end you can
- Package your skills into a clear, specific service offer.
- Run a discovery conversation, scope a project, and price it.
- Communicate professionally and handle a client's data ethically and privately.
- Build credibility in public through a portfolio, case studies, and content.
Weekly plan
| Week | Focus |
|---|---|
| Week 29 | Positioning & offer — who you help, what you deliver, how you price it. |
| Week 30 | Discovery & proposals — the questions to ask, scoping, and writing a proposal. |
| Week 31 | Delivery & ethics — data privacy, setting expectations, professionalism. |
| Week 32 | Build in public & deliver the capstone. |
Free resources
- Freelance-platform guides (Upwork / Fiverr) FREE
- Consulting & proposal basics FREE
- LinkedIn "build in public" guides FREE
Assignments
Service Offer + Portfolio Polish
A one-page service offer ("I build AI automations for small businesses") and a polished portfolio showing all milestones.
Proposal & Pricing
Given a mock client brief, write a real proposal with scope, timeline and price.
Deliver a real AI solution to a real client
Find one real client (free or discounted is fine) and deliver a complete AI solution end-to-end. Produce a case study and a short demo/presentation video. Graded against the full capstone rubric — this is the certificate of the whole course.
Grading & Rubrics
Assessment is standards-based, not a percentage. Weekly assignments are checked simply as done / needs-work (this is formative — practice). What decides whether he advances is the summative milestone at the end of each level, graded on a 4-point scale.
The 4-point mastery scale
The advancement rule: he must score Proficient (3) or higher on a level's milestone before starting the next level. If it's a 1 or 2, he repeats the project — that's not failure, that's how mastery works.
Every milestone is graded on these four dimensions
| Dimension | What "Proficient (3)" looks like |
|---|---|
| Technical execution | The thing actually works, reliably, without someone else fixing it. |
| Problem fit | It solves a real, specific problem for a real, specific person — not a toy demo. |
| Communication | A clear README / case study a non-expert can follow. Explains what, why and how. |
| Ethics & responsibility | Respects privacy, is honest about AI's limits, and verifies outputs rather than blindly trusting them. |
Standards Map
Every level maps to recognised international frameworks — so this isn't a random YouTube playlist, it's a curriculum a school or employer would recognise.
| Level | UNESCO dimension & stage | AI4K12 Big Idea | Industry / Anthropic |
|---|---|---|---|
| L0 Orientation | Human-centred mindset · Ethics — Understand | #5 Societal Impact | AI Fluency intro |
| L1 AI Fluency | AI techniques & applications — Apply | #4 Natural Interaction | 4D Framework · Prompt engineering |
| L2 No-Code | AI system design — Apply → Create | #4 Natural Interaction | No-code automation (n8n/Make) |
| L3 Programming | AI system design — Create | #3 Learning · #2 Representation | Python · LLM APIs |
| L4 RAG & Agents | AI system design — Create | #2 Representation & Reasoning · #1 Perception | RAG · Agents · Evaluation |
| L5 Consulting | Human-centred mindset · Ethics — Create | #5 Societal Impact | Consulting · GTM · Ethics |
Progress Tracker
Tick each milestone as he reaches Proficient. Progress saves automatically in this browser, so he can come back to the same page and see how far he's come.