Your account

Build your own AI agent.

An agent is an AI that does a job instead of just chatting. You build one from an empty Python file, on your own PC, and the AI itself costs nothing.

Start free lesson 0 → or enrol, $49

Lessons 0 to 3 are free. No sign-up, no excuses.

the head (LLM) your code, agent.py (drag the jar to turn it)

The head writes.
The jar does.

  1. The head in the jar up there is an LLM (large language model): a program that read a huge pile of text and predicts what comes next, like your phone’s autocomplete, only far better. It can only write text.

  2. The jar around it is your code, agent.py: the head’s hands. It opens files and runs things for it.

  3. Together they do a job: ask which of your notes mention rent, and the head asks for the files, the jar opens them, and the head reads them and answers. That’s an agent.

Twelve lessons, one file.

Tiny steps: run a few lines, see what happens, then get the next ones. You need some programming, no AI. If you can write a for loop, you’re overqualified for lesson 0. Don’t know Python? Every step shows the exact code and what each line does.

  1. 0.Build an AI from nothing free

    you learn what an AI is and what training means

  2. 1.Make the AI answer from your own code free

    you learn what a prompt is: the text your code sends the AI

  3. 2.Give it a personality free

    you learn what a system prompt is: standing orders it reads before every message

  4. 3.Make it remember the conversation free

    you learn why the AI forgets between messages

the rest comes with the course: being written now, each one opens the day it’s out

  1. 4.Run it out of memory on purpose

    you learn tokens, the pieces the AI counts text in, and why it can only read so much at once

  2. 5.Give it its first tool, reading a file

    a tool is a job your code does for the AI; you learn how the AI asks for one

  3. 6.Let it act in a loop until the job is done

    you learn the agent loop: think, act, look, repeat

  4. 7.Give it more tools

    you learn describing tools so it picks the right one

  5. 8.Let it search your own notes

    you learn embeddings and RAG: finding notes by meaning, not by exact words

  6. 9.Catch it lying

    you learn how to test an AI

  7. 10.Stop it from wrecking your PC

    you learn AI security

  8. 11.Final exam

    your agent solves a real problem of yours, and you explain every piece

think act look repeat until the job is done
The AI replies “read notes.txt”. agent.py opens the file and pastes the text back. The AI reads it and picks the next step, round and round until the job is done: that loop makes it an agent (lesson 6).
“The head only guesses the next word. Everything else, you build.”
- Prof. Viktor Graves

The whole course

$49at launch, then $67

Enrol for $49 → or try lesson 0 first

Makes your account (email or Google), then Stripe takes the payment.

Start free lesson 0 →