New to AI? Start here

How building with AI works

No coding needed — just your ideas and the right words. Poke at every animation below, and you'll walk into AI Builders Day already knowing the ropes.

1

What's an LLM?

“LLM” stands for Large Language Model — the “AI” you'll build with. Here's the whole idea in one picture.

An LLM is a super-smart autocomplete. Nobody typed in the answers — it learned patterns by reading a giant library of writing. It studied millions of sentences like:

The cat sat on the mat.

…so now it's great at guessing what comes next:

The cat sat on the ___
or tap a word yourself

It leans toward its best guess, but it could pick another — that's why it can be creative. Chain thousands of these guesses and it writes stories, answers questions, and builds apps.

2

How do they work?

An LLM answers by predicting the next word again and again. Watch it build a sentence from a prompt.

First the AI reads your whole prompt — every word gives it clues. Then it writes the answer one predicted word at a time.

1 · It reads your prompt
Makemeanapp→ 🧠
2 · It writes the answerword 0/8
Press “Predict next word” →

Every new word is a fresh guess based on everything so far — like stacking LEGO bricks into a wall, one at a time.

3

What are tokens?

Before the AI can read your words, it breaks them into little pieces called tokens. Try splitting some yourself.

Computers don't read whole words — they chop text into little pieces called tokens. On average a token is about 4 letters. Tap a phrase, or type your own:

unbelievable
3 tokens · 12 letters

Under the hood every token is really a number — tap “Show token IDs” to peek. The AI does all its thinking with those numbers, then turns them back into words for you.

4

Credits & prompts

Building costs credits — a bit of “fuel” for each request. See how clear prompts stretch them further.

Credits are like arcade tokens. Every time the AI builds or fixes something, it uses a few. You get a bucket for the day — spend them wisely and they go further.

Credits left100 / 100

Tip: clear, one-thing-at-a-time prompts cost fewer credits and work better.

What uses credits
  • Building or changing your app
  • Fixing a bug you found
  • Longer, more detailed requests
5

Start prompting

A prompt is how you talk to the AI. Compare a vague prompt with a clear one, then try your own follow-ups.

A prompt is you telling the AI what you want. The best ones have three ingredients:

make a game where a cat catches falling pizza on a sunny park background and shows the score at the top

What to buildHow it looksThe goal

See the difference a clear prompt makes — tap one:

🤖 AI response

Tap a prompt above to see what the AI does.

Now build your own prompt
What to build
How it looks
The goal
Clarity:
Pick some ingredients
6

Your step-by-step guide

Stuck on where to begin? Follow these five steps every time you build something new.

Your app so far
🧩 Animal Quiz
7

Publishing with Lovable

Made something awesome? Publishing puts it online so anyone can visit. Give it a try.

When your app works, you publish it. Lovable puts it on the internet and gives you a link (a URL) — send it to anyone and it opens on their phone. Try it:

Your screen
The internet
🔒 You choose who sees it

Nothing is shared until you send the link. You decide who gets it.

🔁 Change it anytime

Keep building and press Publish again — the same link shows your newest version.

Still have questions?

Grown-ups' most-asked questions — about AI, credits, and keeping kids safe — are all answered in the FAQ. Prefer to just ask? Email us at [email protected].

By using our website you agree to our use of cookies to deliver a better user experience. Cookie Policy.
Getting started with AI — AI Builders Day