From Browser to Your Machine

Every tool you've built so far in this track has lived in a browser tab. You type a prompt, Claude responds, you copy the output. That works — and it's genuinely useful. But there's a ceiling.

Browser-based AI can't read the file sitting on your desktop. It can't automatically run every Monday morning. It can't send an email on your behalf without you clicking. It can't access your company's internal systems. To cross any of those lines, your code needs to run on your machine.

This lesson is about what that actually means, and how to set it up without it feeling like a software engineering course.

What "Running Locally" Actually Means

When a program runs locally, your computer's processor executes it. The code can open files on your hard drive. It can make network calls using your internet connection and your credentials. It can be scheduled to run at 3am while you're asleep.

Compare that to what happens in a browser. The browser is a sandboxed container. It can only see what you deliberately give it — a file you upload, text you paste. It cannot browse your file system. It cannot store anything permanently unless you download it.

The gap between "browser-based AI" and "local script" is not a gap in AI capability. It's a gap in what the code is allowed to touch. The same Claude that helps you draft text in a chat window can also write a Python script that runs on your machine and processes a hundred spreadsheets while you sleep — you just need a place for that script to live.

The Three Things You Need

To run code locally, you need three things:

1. Python — the language we'll use throughout this track. Python is free, runs on Windows, Mac, and Linux, and is the most common language for the kind of scripts non-engineers build when they first start automating work.

2. A terminal — the black or white window where you type commands. It's not as scary as it looks. By the next lesson, you'll understand exactly what it is. Mac has one built in (called Terminal). Windows has PowerShell.

3. Claude (your subscription) — to write your code and explain what it does. You use your own Claude account. This academy does not run your code, and it does not access your files. You build locally, Claude helps you build.

That's the full stack. No server. No hosting. No deployment pipeline. Just your machine, Python, and Claude helping you write and understand the scripts.

Setting Up: The Shortest Path

Mac users: Python 3 is installable from python.org or via a tool called Homebrew. Terminal is in your Applications → Utilities folder. For Claude Code (Anthropic's AI-native terminal application), visit claude.com/claude-code and follow the install guide. You'll need a Claude Pro subscription.

Windows users: Download Python 3 from python.org (check "Add Python to PATH" during install). Use PowerShell as your terminal. Claude Code is also available on Windows.

If you just want to start immediately: Open claude.ai in your browser and use it to help you write your scripts. Copy them into a plain text file, save it with a .py extension, and run it from your terminal. That is a completely valid workflow and what most beginners use.

You do not need Claude Code to follow this track. Claude Code is the more powerful, integrated experience — but a Claude.ai chat window plus a terminal is all you need to get started.

The "I Don't Want to Install Anything" Option

If installing Python feels like too much right now, that's fine. Many people start by writing scripts in Claude's chat interface, reading them carefully (which you'll learn to do in "Reading Code You Didn't Write"), and running them when they're ready. The installation step is a ten-minute one-time task, not a gating requirement.

The goal of this lesson is not to get Python installed today. It's to make sure you understand the landscape before the next three lessons ask you to build in it.

A Note on What This Academy Does and Doesn't Do

This is worth saying clearly: the BuildChallenge exercises you'll see later in this track are display-only teaching scaffolds. They show you starter code, explain what each piece does, and give you a reference solution to read. The academy does not execute your code, does not access your machine, and does not store anything you type.

You build with your own Claude subscription, on your own machine. The academy's job is to teach you how to read, understand, and direct that build. That's the partnership.

Where You're Headed

The next lesson demystifies the tools themselves — files, folders, the terminal, what a repository actually is. After that, you'll learn to read code you didn't write (which is what you do with every AI-generated script). Then you'll run your first real script.

By the end of these four lessons, you won't be a developer. You'll be something more useful for your situation: a person who can see a Python script, understand what it does, work with Claude to modify it, and run it on real work.

That's the bridge we're building.