# The AI builder

The AI builder is the heart of GenVibe. You chat with it, and it writes and edits the real code of your app — streaming changes into a live, running preview.

## How it works

1. You send a message describing what you want.
2. The AI generates or edits files, which stream into the **workbench** as it works.
3. The app runs instantly in the **live preview** — no build step, no waiting on a server.
4. You keep chatting to refine, fix, and extend.

## It remembers your project

GenVibe keeps the full context of your app and conversation, so you can make incremental changes without re-explaining everything. Just say what to change next.

## It writes real, standard code

Your app is built with mainstream frameworks and standard code — not a proprietary format. That means:

* You can read and edit every file yourself.
* You can export it to GitHub or deploy it anywhere.
* You're never locked in.

## It can fix its own mistakes

If your app hits a runtime error, GenVibe can detect it and attempt a fix automatically. You can also point out problems yourself:

> The page is blank after I click Submit — please fix it.

## Choosing a model

Use the model menu in the prompt toolbar to choose the AI model. Free accounts use Claude Sonnet; every paid plan (Standard and Pro) unlocks Claude Opus for the most demanding builds.

## Tips for best results

* Describe the whole app in your first message, then iterate in small steps.
* Be specific about behavior and design.
* Use the [inspector](/features/preview-and-workbench.md) to target a specific element.

See [Prompting best practices](/getting-started/prompting-best-practices.md) for the full guide.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.genvibe.pro/features/ai-builder.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.