Five starting points for building with AI

For your first AI-built web page, start with JoCoding’s lottery-number page at 28:17–31:31. You can see the cycle of asking, running, checking and revising.

If you cannot choose a project, start with Lefty Kim. If you cannot find where to type your request, try Coding Noona. Bite-sized IT explains HTML, CSS and JavaScript; Yalco explains why to keep a history of changes.

The linked sources are in Korean. English captions have not been verified. Screenshots preserve the original interfaces; quotations below are English translations of the cited Korean excerpts.

1. What actually happens after you ask AI to build something?

Watch for: how the first result informs the next request.

JoCoding builds a page with a button that produces lottery numbers. He saves the AI-generated code to a file, opens it in a browser and presses the button to check that one set of numbers appears.

After seeing it, he decides that the page should produce five sets at a time. He returns to the conversation and asks:

“

Change it so the button produces five sets of lottery numbers at once.

30:40–30:46, translated excerpt ↗

He saves the revised code and presses the button again. Five sets appear.

Try the same sequence in your project: request one behavior, run it yourself, then describe what to change.

2. If you have too many ideas, where should you start?

Take away: choose something that matters to you and can be attempted this week.

Lefty Kim describes how participants chose their first tasks in AI training at Woowa Brothers:

“

Does this problem really matter to me?

Can I finish it this week?

translated excerpts ↗

Participants were encouraged to choose a problem that scored well on both questions. An important but difficult task was split into smaller pieces. An easy but unimportant one called for a different subject.

The examples are specific: a designer could ask AI to suggest a name and tags for one image; a marketer could begin with a script that retrieves data from one website.

For a personal project, try recording a book title and a one-line note in a list before building a complete reading service. Login and recommendation features can wait until you decide what is needed next.

3. The tool is open. Where do you type the request?

Learn to locate: your project folder and the place where you ask AI to work.

If you have chosen Claude Code, Coding Noona’s setup segment walks through selecting a folder and finding the request input.

She first creates a folder for the project. In VS Code, she opens File → Open Folder and selects it, explaining that the project’s work will be saved there.

Next, she runs claude in the terminal and types a task into the prompt that opens. The terminal is a place to enter commands; whether it appears on the right or below the editor does not change that role.

After installation and login, locate the project folder and request input on your own screen. This segment is about finding where to begin. Check the tool’s current access requirements before following along; your interface may differ.

4. AI says it changed HTML and CSS. What does that mean?

Understand: structure, appearance and behavior have different jobs.

Bite-sized IT explains HTML, CSS and JavaScript through a familiar YouTube interface. At 06:21–06:28, he describes HTML as the language for the arrangement and structure of elements.

He then uses CSS for appearance, such as background colors and fonts, and JavaScript for features that respond to the user, such as search suggestions.

On your page, think of HTML as the structure of elements such as a button, CSS as color, size and spacing, and JavaScript as what happens when it is pressed. These roles do not always live in separate files.

5. How do you go back when a working screen breaks?

Understand: why to record a state that you have checked works.

Yalco introduces Git through a familiar problem: a working feature stops working after an edit. He compares useful recorded states to save points in a game at 03:27–03:31.

Git tracks changes to files. The video explains how recorded intermediate states can help you return to an earlier point if a new version has problems, and why manually undoing a change across several files is difficult.

Before your next edit, learn how to check and record a working state. The selected segment explains why Git is useful; the hands-on recording and recovery exercises come later in the course.

If using AI has made you want to write code yourself, continue with three videos for your first day of coding, from creating a file to checking printed output.