Build your own learning platform. Then use it to learn anything.
A free, practical course for people who don't code. Put an AI agent on your computer, build your own mini Udemy, and learn how to keep improving it yourself.
- Free course
- Small, practical steps
- Starts 10 November 2026
You'll need your own AI agent subscription. No subscription to Prompt It Yourself.
“I'm not technical.”
That's fine. You don't need to be.
The people I've helped on my live show aren't developers either. They're psychologists, accountants, artists, and business owners.
Take Istvan, for example. He had never used an AI coding agent before.
“I don't know what I'm doing but I feel like I'm brave.”
Istvan, projection artist
That's what I want you to discover: what you can do with your own agent.
“Okay, but what exactly am I building?”
Your own learning website. Think of a small Udemy, but made for you.
We start with a course that teaches terminal commands. Yes, we're building a learning platform that teaches you how to use the tools we're building it with.
Very meta.
After that, you can ask your agent to create courses about whatever you're curious about.
Pizza making? Photography? Schoolwork for your kids?
You choose the subject. Your agent helps create the lessons, exercises, and learning experience.
You'll also learn to check the content, because AI can get things wrong.
And if you'd like to share the platform with your family, I'll show you how to put it online.
It's yours. Keep changing it.
What would you learn on your own platform?
Type anything. This is what your agent could draft first. Then you check it.
Your agent drafts it. You check it, because AI can get things wrong.
“So what am I actually learning?”
How to work with your AI agent.
We go through small, practical steps, from installing your agent to building your own learning platform.
Along the way, you'll practice how to:
- Explain what you want.
- Give your agent useful context.
- Ask it to make a plan.
- Challenge ideas and request changes.
- Test the result and investigate mistakes.
- Save your work and keep improving it.
Every new concept starts with a short explanation. Then we do the actual work.
I'll show you what I do. Then it's your turn.
The agent writes the code. You make the decisions.
“Half the words you're using sound like gibberish.”
Fair.
That's why I've created short explanations, presentations, and animations to introduce each concept.
Then we practice with real examples.
You will encounter unfamiliar words. You're learning a new skill, after all.
And here's something useful you can do immediately: when your agent says something you don't understand, ask it to explain in simpler language.
You don't have to pretend you understood the first time.
“I already pay for ChatGPT. Why are you using Claude?”
Keep your ChatGPT subscription if you like it.
ChatGPT has coding-agent capabilities too, and other agents can do similar work.
I demonstrate Claude Code because that's what I use. The installation instructions, screenshots, and exercises will match Claude.
If you choose another coding agent, you'll need to adapt some steps.
But the important skill (learning how to communicate, collaborate, and verify) can transfer between agents.
“What if I don't like what my agent builds?”
Tell it.
Ask for another version. Change the colors. Move something around. Compare a few ideas.
Istvan said something I love:
“I could never dare to ask 10 variations from a human designer.”
Istvan, projection artist
And here's another idea: instead of requesting one change after another, ask your agent to build you a little playground where you can experiment yourself.
You don't have to accept the first result. Ask for a rough version first, and let your agent tell you where it guessed.
And you don't have to know the perfect design before you begin.
“So you're going to share your code?”
No. I share the process, not my code.
Your project might look completely different from mine.
Good.
I don't want you to memorize my solution. I want you to learn how to build and change yours.
If something doesn't work, we'll practice asking your agent to investigate and explain the problem.
“Can I use this knowledge for anything else? Even at work?”
I think so.
The learning platform is our first project. The method can be useful for many other things.
Adrienn is an accountant. A few weeks after we started working together, she came back on my show.
One of the first things she said was:
“since then, I created some apps.”
Adrienn, accountant
See what Adrienn built independently
That's what excites me.
She didn't wait for our next session to start experimenting.
If you use these skills at work, make sure you follow your employer's AI and data-security policies.
“What if I get stuck?”
You probably will. Haha.
I still do.
In fact, one of the first lessons in Fundamentals is called "Stuck? Show it and ask."
Something doesn't work? Show your agent the error. Explain what you expected to happen. Ask what you should try next.
Don't understand the answer? Ask it to explain.
Still not working? Ask it to investigate, check the documentation, or try a smaller experiment.
I've already made Shorts about this exact problem:
And if your agent seems to have forgotten everything: it didn't. You probably closed the terminal.
I share practical ideas like these regularly. One problem, one thing you can try.
And you can join my free Sunday group webinars to ask questions.
I can't personally debug every project, and your agent won't always have the answer either.
But you'll learn how to investigate problems and find a way forward.
That's a skill worth having.
“Hang on. What can this agent actually see on my computer?”
Good question.
That depends on the agent, its permissions, and how you set it up.
We'll talk about working in project folders, understanding permissions, protecting your files, and checking what the agent wants to do.
Don't give an agent unlimited access to sensitive information just because it asks.
And for this course, use your own Mac or Windows computer, not a work computer. Companies usually put guardrails on theirs, so the setup may not work there anyway.
One useful habit is to start small and limit what can go wrong.
“Wait. Is this really free?”
Yes.
Fundamentals is free. No trial, no course fee, and no credit card required.
But you do need your own AI coding agent.
I demonstrate Claude Code. Claude Pro is enough to get started, and it's a paid subscription to Anthropic, not to me.
You might already have access to another coding agent through a subscription you use. Just remember that the course demonstrations use Claude.
You'll also need your own Mac or Windows computer.
Your learning platform can run on your computer. Putting it online is part of the project, and some choices, such as buying a domain, may cost extra.
“So I have to register?”
Yes.
Your name and email address. I'll email you a sign-in link, so you don't even need another password.
The course opens November 10, 2026, and will be available in your account.
You can also join the free Sunday webinars and report problems in the lessons.
The newsletter is optional.
No credit card.
“Still sounds fishy. How do you make money?”
Fair question.
I'll create other standalone courses and case studies based on my own experiments.
Video production, marketing, building useful tools, and whatever interesting problem I work on next.
Those will be optional purchases.
I may also create extra lessons showing how to add features to the learning platform, such as a better knowledge base or a new interaction.
But Fundamentals is complete without buying those extras.
I also offer private sessions and may run workshops for businesses.
You don't need to keep paying me to continue using what you build.
“And who's sponsoring you?”
I wish! Haha.
Nobody is paying me to recommend Claude Code, and I have a no-affiliate-links policy for external services.
I'm not against commercial tools. I use plenty of them.
But I believe you should start with what you actually need instead of collecting subscriptions and complicated tools.
Your agent can help you research alternatives and decide what makes sense for you. It can even check what a paid AI service really does.
“So I'm just asking AI to do the work. Won't that make me dumb?”
I don't think so.
If you accept every answer without thinking, you're probably not learning much.
But if you're asking questions, comparing ideas, challenging results, and making decisions, you're still doing plenty of thinking.
Keep the part only you can do. Hand over the research, the lists, the writing and the sorting.
I've spent 25 years working in IT. That doesn't make me an experienced designer, video editor, or marketer.
Yet I've been able to experiment in all those areas with my agent.
I don't think I'm becoming less capable.
I'm getting more done and learning things I wouldn't have attempted before.
And there's one thing you should know.
You're going to have to read. Quite a lot, actually.
Your agent can write the code. But you still have to understand what you're asking it to do, read its explanations, and check the results.
If you're looking for a magic button that does everything while you sit back, this probably isn't your course.
If you're curious, willing to experiment, and ready to learn something new, welcome.
Start learning. Prompt it yourself.
Build your first learning platform. Learn to work with your agent. Then see what you can create next.
Fundamentals is free and starts 10 November 2026. You need your own AI agent subscription.