Home / Learn / Data & AI / Offline AI / Lesson 1.1
All lessons

Module 1: Welcome, and why run AI yourself · Lesson 1.1 · about 15 min

What this course gives you

A map of the whole course, the four things you will build, and how to study it so the hands-on parts work first time.

Key points, meanings and formulas on one page: choose "Save as PDF" in the print box.

Module 1 · about 15 minutes · read first: nothing. By the end you will know what the course covers, what you will build, and which lessons to take first.

This course shows you, one short lesson at a time, how to download an AI model and run it on a computer you control, then put it to real work.

1. Why this matters

Most people meet AI through a website. You type, a company’s computers think, and an answer comes back. That works until something gets in the way: private files you cannot upload, no internet, a monthly limit you keep hitting, or a job that asks you to build AI into a workflow.

This course teaches the other way. You download a model and run it on a machine you control. Think of it as the difference between eating at a restaurant and cooking at home: the restaurant is easy, but at home you choose the ingredients, nobody sees your recipe, and the kitchen is open at 3 a.m.

2. What you will be able to do

By the end of the course you will be able to:

  • explain in plain words what a model is, what a token is, and why memory decides what you can run;
  • choose hardware for your budget in India, Canada, the USA or the UK, and know what it costs to run;
  • download a model such as Google’s Gemma 4, install it, and use it with the internet switched off;
  • call it from small Python programs, so it can work through your files while you do something else;
  • search your own documents with it, privately;
  • share one machine with a family or a team;
  • compare the real cost with Claude, ChatGPT and Google plans, and know where each wins.
Course map
  1. A
    Foundations
    What a model is, tokens, and why run AI yourself.
  2. B
    Hardware
    Memory, speed, power, and what to buy in your country.
  3. C
    Software and setup
    Download a model, shrink it to fit, install and tune it.
  4. D
    Offline pipelines
    Connect it to Python, your documents, speech and agents.
  5. E
    Scale, money, limits
    Serve a team, compare costs, know the limits, find your sector.
You leave with:
  • A private assistant
  • Search your own documents
  • Your own workflows in Python
  • A cost sheet against subscriptions
Each part builds on the one before. The hands-on laptop path (install a model, then call it from Python) comes early, so you can start using local AI after Part C.

3. How the course is built

There are 5 parts, 18 modules and 119 lessons. Every lesson follows the same shape:

  1. A big idea in one sentence, so you know where you are going.
  2. Why this matters, in two to four sentences.
  3. Numbered sections, each a plain explanation first, with an everyday picture.
  4. Watch it happen: a short step-by-step player. Press Play, or step through with Next.
  5. A meaning box that says what each symbol stands for and what it means in the real world.
  6. A “Maths and principle” box with the rule behind it, one equation, an analogy and a tiny calculation. You can skip the box and still follow the lesson.
  7. A worked example, and live problems that make new numbers each time you press the button.
  8. Practice, with answers hidden until you open them, and key points you can download as a one-page summary sheet.
  9. Check yourself: three quick questions at the end.

Some lessons have simulations: small things to play with, such as a slider that shows whether a model fits in your memory. They run in your browser and send nothing anywhere. Nothing about you is stored.

Here is how one study plan works out, with the numbers from the worked example below.

Count the lessons. The short path (section 4) is about 25 lessons.

Time per lesson. About 20 minutes of reading and 10 of practice: 30 minutes. So 25 × 30 = 750 minutes, which is 12.5 hours.

Your daily time. Priya has 40 minutes at lunch on 5 days a week: 40 × 5 = 200 minutes a week.

Divide. 750 ÷ 200 = 3.75 weeks. In about four weeks she has a model running offline and a script doing real work.

Maths and principle

Principle: learning sticks when you do something with it within a day of reading it.

Equation: total time ≈ lessons × (reading + practice) = 119 × (20 + 10) minutes = 3,570 minutes, about 60 hours.

Analogy: the course is a sum of small terms. Each lesson adds a little; skipping the practice is like dropping terms from the sum and hoping the total stays the same. It does not.

Try it: at 45 minutes a day, about how many weeks is 60 hours? (60 × 60 ÷ 45 = 80 days, about 11 to 12 weeks.)

L
the number of lessons you plan to take
r + p
minutes of reading plus minutes of practice for one lesson, about 20 + 10
T = L × (r + p)
the total study time in minutes
Physical meaning
The course is long, but each piece is small. At half an hour a day the short path takes about a month and the whole course about a season, the same as learning to drive.

4. Which order to take it in

The course is written to be taken in order, but you do not have to. If your first goal is “get a model running on my laptop and use it from code”, take this short path first:

  1. Module 1 (this module) and Module 3, lessons 3.1 to 3.5 (tokens).
  2. Lesson 4.1 (memory), 7.1 and 7.3 (finding and choosing a model), and 8.1 and 8.5 (shrinking it to fit).
  3. Module 9 (install, then test with the internet off).
  4. Module 11 (call the model from Python).

That path is about 25 lessons, all written. After it you can use local AI for real work, and the rest of the course makes you better at it. Lesson 1.4 has a short quiz that suggests a path for your own situation.

“I have to read every lesson in order before I can do anything useful.” You don’t. The short path gets a model working within about a month of lunch breaks. The other lessons explain why it works and how to make it faster, cheaper and safer, and you can come back to them whenever a question comes up.

5. Worked example

Made-up numbers. Priya works in a small accounts office. She has a laptop with 16 GB of memory and about 40 minutes at lunch. She takes one lesson a day. After about four weeks of the short path she has a model running offline and a Python script that summarises each client’s PDF statements into a one-page note. She has spent nothing on hardware. She then reads Module 16 to decide whether a desktop would pay for itself.

6. Practice

  1. Write down the one job you most want local AI to do for you.
  2. Find out how much memory (RAM) your computer has. On Windows: Settings, System, About. On a Mac: Apple menu, About This Mac.
  3. Decide whether you will take the course in order or the short path first.
Answers and tips
  1. Keep it concrete: “summarise my meeting notes” beats “use AI more”. You will come back to this in the capstone.
  2. Write the number down. 8 GB, 16 GB or 32 GB decides which models you can run, as Module 4 explains.
  3. If you want results this month, take the short path; if you want the full picture first, go in order.

New words: model (the file that holds what the AI learned), local or on-device (runs on your own machine), cloud (runs on a company’s computers over the internet).

  • Local AI means the model file sits on your machine and the work happens there, not on a company’s servers.
  • The course has 5 parts, 18 modules and 119 lessons; each lesson is about 20 minutes of reading and 10 of practice.
  • The short path (about 25 lessons) gets a model running offline and called from Python in about a month.
  • Total time = lessons × (reading + practice); one lesson a day with its practice is a good pace.
  • Do the practice: it is what makes the setup work first time.

Check yourself

1Which part of the course gets a model running on your own machine?
2What does "local AI" mean in this course?
3You have 30 minutes a day. What is a sensible pace?

Sources

Your ticks stay in this browser. Save them as a file on the course home.

Updated 2026-11-22. Prices and model names change fast: check the official source before you buy.