Summary  
This chapter covers prompt engineering techniques for AI coding assistants, emphasizing the importance of writing literal, narrowly scoped prompts and following an iterative write–review–adjust loop to make precise code modifications.

General domain of usage  
AI-assisted software development

You have a working app. Now you need to know how to talk to Claude Code effectively so that when you want to change something, it changes exactly that thing and nothing else.

### The build loop

Every change you make from here follows the same pattern.

```
Write a prompt → Review the output → Adjust if needed → Repeat
```

One thing at a time. One prompt at a time. That is it.

Say exactly what to change and how, no vague descriptions.

Be literal

Specify what to change and what not to touch.

Limit the scope

Turn rough ideas into clear, specific prompts before using them.

Use AI to refine prompts

Update the target audience field on the idea form to be a dropdown with preset options. Do not change anything else.

Add a selection wheel to the target audience field with options for freelancers, SMB owners, and college students. Keep the field optional.

The more specific your prompt, the more predictable the output. Vague prompts produce vague results. Specific prompts produce exactly what you asked for.


Note

Which approach leads to the most predictable results when prompting an AI to update your app?

Build and launch a real app in just one weekend using modern AI tools and a simple, practical workflow, using Claude Code to plan, generate, and refine your code, managing your project with GitHub, and deploying your finished app online with Vercel. You’ll start with an idea and quickly turn it into a clear project plan, then move step by step into building a working application, focusing on core features, simple architecture, and fast iteration instead of unnecessary complexity, while learning how to work effectively with AI by writing better prompts, fixing errors, and improving features, ending with a live app, a clean GitHub project, and a repeatable system for turning future ideas into real products faster.

Understand the basics of AI-assisted development, set up your environment, and get familiar with Claude Code before starting your first app.

Turn your idea into a clear, simple plan and define the smallest version of your app that delivers real value.

Go from zero to a running app by generating code with AI, setting up the project, and understanding what was created.

Improve your app by adding features, refining the design, and learning how to fix issues effectively.

Publish your project, make it accessible online, and ensure everything works using GitHub and Vercel.

Writing Better Prompts

The build loop