Learn AI by playing

You do not need a forty-hour course to understand how artificial intelligence works. You need a little practice every day, room to be wrong without penalty, and the why behind every answer. That is what a good game does.

Why playing works for learning AI

AI makes more sense when you watch it fail than when you read definitions. When you see an assistant make up a fact, forget what you told it or overspend, and then you fix it yourself, the idea sticks. Read in a handout, it slips away.

A well-built game has exactly what that kind of learning needs:

Predict before you see. Committing to an answer, even a wrong one, makes you pay attention to the result.
Instant feedback. You know right away whether you got it, and why.
Recall, do not reread. Answering questions about what you learned locks it in better than reading it again. It is one of the best-supported study techniques.
Spaced review. What you missed comes back later, right when you are about to forget it.
Short sessions. Five minutes a day is sustainable; a Sunday marathon is not.

What to understand first

There are lots of new words around AI, but almost everything rests on a few ideas. In this order, each one sets up the next (they are all in our AI glossary):

What a token is: the unit the model reads, writes and bills you in.
What an LLM is: a model that predicts the next token, and why that is enough to hold a conversation.
The prompt and the context window: what you give the model and how much it can keep in mind.
Hallucinations and temperature: why it sometimes makes things up and what randomness has to do with it.
RAG and embeddings: how you give it your own, up-to-date information.
Agents and MCP: when the model stops just answering and starts using tools.

How Tokencito plays

Tokencito is the game we made for this. At its center is a small system that misbehaves: an assistant that makes up answers, forgets what you told it or overspends. You save it in four steps:

Predict: before running it, pick what you think will happen.
Run: watch it answer and find out if you were right.
Fix: change its instructions and watch it change in front of you.
Apply: solve the same problem in a new context.

A few minutes a day

To keep the habit going, Tokencito adds short pieces you can play in a couple of minutes:

Game of the day: up to five questions, the same for everyone.
Daily streak: every day you play adds up, and if it breaks you start again without losing anything.
Review that comes back: what you miss or save returns tomorrow, in 3 days and in a week.
Guess the token: pick the word that comes next and see how a model decides. It is the most direct way to get what a token is.

Three paths, depending on where you start

Artificial intelligence, no code: for people who do not code. What a token is, how a model predicts, what it costs, why it makes things up and how to fix it.
Agents, for people who code: coding agents, subagents, skills and MCP, what they really do and what they only say they did.
AI costs: why one model costs more than another, what effort is, and how to pay less without losing quality.

Frequently asked questions

Can you learn AI by playing?

You can get a solid grasp of its core ideas: what a token is, how a model predicts, why it makes things up and how to correct it. A game will not make you a machine learning specialist, but it gives you the basis to use AI with good judgment and to keep learning.

Do I need to know how to code?

No. Tokencito's artificial intelligence course needs no code. The agents course is meant for people who code.

How much time a day does it take?

Five minutes is enough. The game of the day has up to five questions, and spaced review brings back what you missed.

Is Tokencito free?

The app is a free download and the basics are free forever: the first path of each course, the game of the day, your streak and review. Tokencito Complete opens everything else with a one-time purchase, no subscription. On the web you play the first session of each course.

Which languages and devices does it support?

English, Spanish and Portuguese. It is on the App Store for iPhone and iPad (it works on Apple TV too), on Google Play, on the Microsoft Store and in the browser.