Your Code is Just Files

It's Just a Folder on a Computer

When engineers talk about "a codebase" or "a project," it sounds technical. It's not. A codebase is just a collection of files organized in folders — the same way you organize files on your own computer, in Google Drive, or in SharePoint.

You already know what this looks like: folders with names, files inside them, some files are documents, some are images, some are data. A software project is the same thing. The only difference is that some of the files contain code — and you don't need to read or write those files yourself. Your AI coding assistant does that for you.

The Safety Net: Version History

Here's something that should make you feel brave: every change to your project is tracked.

Imagine a shared document that remembers every edit anyone ever made — who changed what, when, and why. You can see every version, compare them, and undo any change at any time.

That's what version history does, using a tool called Git. If you've ever opened "Version History" in Google Docs or tracked changes in Microsoft Word, you've seen a simpler version of this — Git is the same idea, built for software projects. You don't need to learn Git commands — your AI coding assistant handles that for you. What matters is knowing this safety net exists:

• You can always undo any change — including ones AI makes
• You can experiment freely without fear of breaking things permanently
• Multiple people on your team can work on the project at the same time

This is why moving from chat to a real project is such a big deal. In chat, if something goes wrong, you start over. In a project with version history, you can go back to an old version.

Save Early, Save Often

There's one catch: version history only protects work you've saved.

Remember from Coder Setup — your project lives on a computer in the cloud, not on your laptop. That cloud computer is reliable, but like any computer, things can occasionally go wrong. If they do and you haven't saved, your recent work could be lost.

The good news: protecting your work is simple, and it has two parts:

1. Save — records your changes. Think of it like hitting Ctrl+S on a Word document.
2. Sync — copies your saved work to GitLab, a safe place outside of Coder. Think of it like moving a file from your computer into Google Drive or SharePoint — even if your computer has a problem, the file is still there in the cloud.

You want to do both every time. Just tell your AI coding assistant:

Save my progress and sync it.

That's it. Your AI assistant takes care of the rest. Even if Coder restarts or your workspace has a problem, your synced work is safely stored in GitLab, ready for you to pick back up.

When to save and sync: Every time you finish something that works — a new feature, a fix, a change you're happy with — tell your AI to save and sync. You'll build this habit throughout the day, and it'll become second nature fast.

This isn't just about backup — it's about building a safety net for experimentation. When you save after something works, you create a known-good point you can always return to. That means you can ask AI to try something ambitious on the next step, knowing that if it breaks, you haven't lost what was already working.

Your AI Coding Assistant

You already know how to work with AI — you've been doing it all morning. Your AI coding assistant is the same kind of capable colleague, with one critical difference: it lives inside your project.

Instead of generating text in a chat window, it: - Reads files in your project to understand what's already there - Creates new files when you need them - Edits existing files to make changes you describe - Runs commands to test, build, and manage your project

You are not writing code. You are directing an AI that writes code on your behalf.

And here's the part that matters most: nothing changes about how you talk to it. The Three Pillars still work. User stories with acceptance criteria still work. The Explore → Plan → Implement → Verify workflow still works. Everything you practiced in Lift 1 applies exactly the same way here. This section might feel more technical because we're talking about files and version history, but the way you actually build things — plain English prompts, clear stories, specific acceptance criteria — is identical to what you've already been doing.

Try It: Change & Save

Format: Pairs Setup: Work with a partner from your team. One person drives (types), the other navigates (watches and suggests).

This exercise walks through two things: making your first change and saving it. By the end, you'll have done everything you need to know to start building.

Open your AI coding assistant by clicking the Terminal button in your Coder workspace. You'll be dropped right into a conversation with your AI assistant — no setup needed.

Step 1 — Make a change: Your project already has files in it — a web application framework is set up and ready to go. There's also a README file. A README is a simple document that describes what a project is — think of it as the cover page. Right now, it's generic. Let's make it yours. Ask your AI coding assistant:

Edit the README file to add our team name. We're the [your team name] team and we're building an Avalanche Field Guide.

Watch what happens. It reads the existing file, makes the change, and saves it — done by talking, not typing code. That change is real. It's not inside a chat window. It's on disk inside of Coder.

Step 2 — Save and sync: Now protect your work:

Save my progress and sync it.

Your AI assistant will save the change and sync it to GitLab. You've just completed the full cycle: make a change, save and sync your work.

After you try it: Did it work? Compare notes with your partner. You both just edited a real file in a real software project — by talking, not coding.

Key Insight

A codebase is just files in folders — the same kind you already work with every day. Version history means you can always undo any change — as long as you save and sync. Get in the habit of telling your AI assistant to "save my progress and sync it" every time you finish something that works. And remember: your AI coding assistant is the same kind of partner you've been working with all morning — it just lives inside your project now, reading and writing real files instead of generating text in a chat window. And because it can now edit real files across your whole project — not just generate text in a chat window — the Explore → Plan → Implement → Verify workflow matters even more. More power means more reason to plan before you build.