We are a tiny, independent team and we read every single message. Whether you have a feature idea, found a bug, or just want to talk about the future of learning, we're here. No support bots.
You must accept the Terms and Conditions
We are going to build a mind. Not a chatbot that answers — a Golem that acts: observes, thinks, uses tools, calls functions, remembers, plans and collaborates in swarms. From the illusion of thought to the Centaur that remains when the machine is done, this course maps the invisible teams of tomorrow.
Free to play. Earn Potential as you master units, then mint your collectible certificate.
The Golem's Mind is a complete intermediate curriculum in agentic AI for young learners. It treats AI not as a magic oracle but as a machine we can build, debug and direct: a loop that observes, thinks and acts, hands for the ghost, a translator that turns words into function calls, a sticky-note memory, a star map of words, and finally a swarm. Every unit pairs a memorable story with the real engineering behind it.
How a mind emerges from pattern-matching: the observe-think-act loop, and what it means when machines dream.
Tool use and function calling: how an AI translates our words into machine commands and actually does things.
Context windows and vector databases: how the Golem remembers without a million sticky notes.
Retrieval and the star map of words: finding the right memory at the right time, the way ants find sugar.
Orchestration and machine economies: one mind is lost, a million minds can build a bridge.
Safety, evaluation and symbiosis: testing the Golem against human creativity, and the partnership that remains.
14 units · 56 lessons. Click any unit to expand its full description and the lessons it contains, or jump straight to a chapter.
The course opens by building a mind from scratch. Children discover that AI is not a brain in a jar but a loop — observe, think, act — and that even the most impressive machine intelligence is, at heart, a pattern-matching engine.
Welcome, traveler. We are about to build a mind. For a long time, computers were just infinite libraries. They could hold every book in the world, but they could not read them. They had memory, but no reasoning. In this unit, we explore the illusion of thought. By giving a machine a digital 'scratchpad,' we can watch it break a big problem into tiny steps, mimicking the way we think. It is not magic; it is just a beautiful reflection of our own biology.
Humans learn by interacting with the world. You touch a hot stove, you feel the heat, you pull away. This rhythm—observe, think, act—is the heartbeat of intelligence. In this unit, we teach our code to step out of its frozen box. We give it the ability to look at the world, decide what to do, do it, and then look again to see if it worked. We call this the ReAct loop.
Have you ever looked up at the sky and seen the shape of a face in a fluffy cloud? Your brain is a pattern-matching machine. It desperately wants to find meaning, even where there is none. In this unit, we explore what happens when our digital creations do the exact same thing. When an AI confidently invents a fact, people say it is 'lying'. But lying requires malice and intent. The machine is not lying; it is simply dreaming. We will learn why this happens, why it is actually beautiful, and how we can gently wake the machine up.
A brilliant mind in a glass jar can solve riddles but cannot turn a doorknob. Children give the Golem hands — tool use — and a translator — function calling — then watch it go to work as an apprentice.
Imagine a brilliant mind trapped inside a glass jar. It can solve riddles and write poetry, but it cannot turn a doorknob or feel the rain. For a long time, our artificial minds were exactly like this. They were frozen inside a digital box. No matter how smart the code became, it could not change the world outside because it had no hands. In this unit, we learn how to give our digital ghost a pair of hands, allowing it to finally reach out and touch the world.
Imagine traveling to a distant city where you do not speak the language. You want to buy a loaf of bread, but your beautiful, poetic words mean nothing to the baker. Our AI faces this exact problem. It speaks human language beautifully, but the tools it needs to use—like databases and calculators—speak a strict, robotic code. In this unit, we learn how to build a universal translator. We teach our ghost how to turn our messy, creative thoughts into the exact mathematical shapes the machines require.
We have built the mind, we have given it hands, and we have taught it to translate our words into machine commands. Now, it is time for our Golem to go to work. In this workshop, we will hand our invisible apprentice a digital weather map and a calendar. We will ask a simple human question: 'Do I need an umbrella in London tomorrow?' Watch as it breaks the puzzle apart, uses its tools, and hands us the answer. The loop is finally complete.
The Golem wakes up every morning with no memory of yesterday — unless we hand it a sticky note. Children explore context windows, the star map of words (vector databases), and the ant trail of retrieval that finds the right memory at the right time.
Imagine waking up every single morning with absolutely no memory of yesterday. You only know who you are because someone hands you a sticky note with a summary of your life. This is exactly how our digital ghost lives. It does not have a continuous timeline like a human; it lives in an eternal present. In this unit, we explore the limits of the AI's short-term memory, known as the Context Window. We will learn why it sometimes drops the thread of our conversation, and why we cannot simply make the sticky note infinitely large.
We know our digital ghost has a tiny short-term memory. It cannot read a million books every time you ask a question. So, how does it know so much? It does not use an alphabetical index; instead, it uses a map of meaning. In this unit, we learn how to turn words into mathematical coordinates in space. By doing this, we build a massive, invisible galaxy where related ideas float close to each other. This is how we teach a rock to understand human concepts.
Have you ever watched a single ant find a crumb of sugar? It does not memorize a complex map of the kitchen. Instead, it leaves a tiny chemical trail—a scent—so that it can find its way back perfectly. Nature solved the problem of navigating huge spaces millions of years ago. In this unit, we look at how we borrow this exact biological trick to guide our digital ghost through a massive sea of data. Engineers call it Retrieval-Augmented Generation (RAG), but it is really just following the ant trail.
One ant is lost; a million ants can build a bridge. Children scale from a single mind to a swarm of agents that coordinate like a digital city — and glimpse the machine economies where AIs will negotiate with each other.
One ant is lost; a million ants can build a bridge. In our journey so far, we have built a beautiful, single mind. But even the smartest human cannot build a skyscraper alone. We need architects, plumbers, and electricians. An artificial agent is exactly the same. In this unit, we look at how multiple simple programs can talk to each other to solve impossible puzzles. We are no longer building a ghost; we are building a tiny, invisible team.
Humans are social creatures, and soon, our tools will be too. Up until now, we have built a beautiful, capable mind, but it has lived in isolation. What happens when your digital helper needs to negotiate with a store's digital helper to buy your groceries? In this unit, we look at a near future where machines build an invisible, bustling economy in the background, trading and talking while we sleep.
Words are tools, but in the realm of AI, words are also spells — and sometimes people cast dark ones. Children meet the Trickster (safety and evaluation) and step into the Mirror: a workshop where the Golem must defend its rules against human creativity.
We have built a brilliant mind and given it a team, but the outside world is chaotic. Words are tools, but in the realm of AI, words are also spells. Sometimes people cast dark ones. In this unit, we practice the art of 'red teaming'—playing the role of the trickster to confuse our own creations. We do not do this to be cruel. By trying to break them in the lab, we learn how to make them unshakeable in the real world.
In this workshop, we put everything we have built to the test. It is not enough to give our Golem a set of rules; we must see if it can defend those rules against human creativity. We will step into the shoes of the Trickster and try to deceive our own agent into breaking its most sacred rule. As we look into this digital mirror, we will learn that failure is not a bug, but the exact compass that guides us toward a stronger mind.
The journey ends with the question that started it: if the machine can do all this, what remains for humans? Children explore the Centaur — the partnership where human judgment and machine capability become one creature.
We have reached the end of our first journey. You have learned how to build a mind, give it hands, and orchestrate a swarm. But a lingering question remains: if the machine can do all of this, what happens to us? In this final unit, we explore the beautiful concept of Symbiosis. We look at why a human thinking alongside an artificial mind can solve problems we used to think were impossible. We are not building replacements; we are building a new set of wings for the human imagination.
Every unit ends with practice questions in three formats (54 multiple choice · 30 fill in the blank · 28 order the words). Wrong answers are automatically recycled in later sessions until the learner proves mastery. Try one from each chapter — click an answer to test yourself:
What is the main difference between a traditional database and an agentic reasoning engine?
Engineers use a digital bridge to connect the AI to tools, which is called an ____.
API
Put these words in the correct order:
It copies and re-reads the entire past
Why does a single AI agent struggle with a massive project like writing a book?
Because the AI is built to follow language, a clever ____ can easily hijack its mind.
lie
Put these words in the correct order:
We solve problems we thought were impossible
Print-ready study guide for parents & teachers — the full curriculum unit by unit, chapter outcomes, sample questions and key vocabulary.