From Chat to Code
0 of 13 lessons done

Your self-paced path from "I have a Claude account I barely touch" to "Claude is the first thing I open when I have something to do." Go at your own speed. Nobody is waiting on you, and you can't fall behind.

How this works

Three sections that build on each other: co-work, then code on the desktop, then code in the terminal. New to all of it? Start at the top and go in order.

We won't assign you a project. Bring something you want to make, or bring nothing and pick from a menu of real tasks when you're ready. Both work.

Start with co-work

Claude as a coworker, inside the desktop app. No code, no terminal. This is where everyone should begin.

Co-work 1Meet your AI coworker

Most people use Claude like a search box. You type a question, you get an answer, you copy the good part out, you move on. That works, and it's also about 10% of what it can do.

Co-work is different. In the Claude desktop app, you can give Claude access to a folder on your computer, and now it's not answering questions in a vacuum. It can read your files, work with what's actually there, and hand things back to you. It stops being a clever stranger and starts being a coworker who knows what you're working on.

The shift to hold onto: you're not writing perfect prompts. You're talking to a capable colleague who happens to be fast, tireless, and a little too eager. You explain what you want the way you'd explain it to a smart new hire on their first day. That's the whole skill.
Mark complete when: you get the difference between asking Claude a question and working alongside Claude.
Co-work 2Get set up

This is the only fiddly lesson. Take it slow and you'll be fine.

  1. Download the Claude desktop app. Go to claude.ai/download and install it for your computer (Mac or Windows). The desktop app is what unlocks co-work. The browser version won't do this part.
  2. Sign in and make sure you're on a paid plan. Open the app, sign in, and check that you're on Claude Pro or higher. If you're on the free plan, upgrade now. Everything from here needs it.
  3. Make a folder to work in. Create one new, empty folder somewhere easy to find, like your Desktop. Call it something like claude-sandbox. This is your playground, separate from anything important.
  4. Connect the folder to Claude. In the desktop app, give Claude access to that folder. The app walks you through this the first time. When it's connected, Claude can see and work inside it.
Mark complete when: the desktop app is installed, you're on a paid plan, and Claude has access to your sandbox folder.
Watch out for: downloading from a copycat site. Make sure you're at claude.ai. The real app comes from Anthropic.
Co-work 3Your first conversation

Don't overthink this one. The goal is just to feel what it's like to hand something off. Open a conversation and talk to Claude the way you'd talk to a person. Out loud is fine, if you use voice. Messy is fine. You don't need the perfect words.

Pick one small task to start with. Something real, something low-stakes, something that would genuinely save you ten minutes. Here are starters if nothing comes to mind:

Quick wins to pick from

Give Claude the task, then talk with it. If the first answer isn't right, say so plainly: "that's too formal," or "you missed the deadline," or "shorter." It adjusts. That back-and-forth is the actual work, and it's how you'd manage any good coworker.

Mark complete when: you've finished one real task by talking it through, including at least one round of "not quite, try again."
Co-work 4Let it into your files

In the last lesson you mostly pasted things in. Now let Claude work with what's already in your folder, because that's where it gets genuinely useful.

Drop a couple of real files into your sandbox folder. Nothing precious. A document, a spreadsheet, a few notes. Then ask Claude to actually use them:

Watch what happens. Claude reads the file, works with the real content, and hands it back. You didn't copy and paste anything. This is the moment most people feel the click, when it stops being a chatbot and starts being someone who knows your stuff.

One habit worth building now: be specific about which file and what you want done to it. "Summarize the meeting notes from Tuesday" beats "summarize this." Specific in, useful out.
Mark complete when: Claude has read at least one real file from your folder and done something useful with it.
Co-work 5Build one real thing in co-work

Time to finish something you keep. Not an exercise, a real output that exists when you're done. Pick one task and take it all the way through, with Claude doing the heavy lifting and you steering.

Finish something real

Work it like a project. Hand it off, react to drafts, push back, refine, until it's actually good. When you've got something you'd genuinely use, you've done co-work for real. Everything in the next two sections is the same skill, pointed at more powerful tools.

Mark complete when: you have one finished thing you'd actually use, built by working alongside Claude.

Code on the desktop app

Same coworker, more powerful tool. Now Claude builds real, working things, and you still never touch a terminal. Do the co-work section first.

Code 1What "code" adds, and when co-work isn't enough

In the last section, Claude worked with documents and words. That's co-work. It's powerful, and for a lot of what you do, it's all you need.

Code is the next door. Instead of handing you a document, Claude builds something that runs. A webpage. A little tool. A page that does something when you click it. You don't write the code, and you don't need to read it. You describe what you want, Claude builds it, and you look at the result and say what to change.

The honest version of when you reach for this: co-work is for thinking, writing, and organizing. Code is for making a thing that exists and works on its own. If you've ever thought "I wish I had a simple page for this" or "I wish there were a little tool that just did X," that's a code task.

The good news: it's the exact same skill you already built. You talk to a coworker. The coworker just happens to build webpages now.

Mark complete when: you can tell the difference between a co-work task (words, documents) and a code task (a thing that runs).
Code 2Turn on code in the desktop app
A note on the steps: the desktop app changes its buttons and labels now and then. If a label below doesn't match exactly what you see, look for the closest thing. The idea stays the same even when the wording moves.

You already have everything installed from the co-work section. This is about pointing Claude at a building job instead of a writing job.

  1. Make a fresh folder for building. Like your sandbox, but for projects. Call it something like claude-builds. Each thing you make gets its own home, so nothing gets tangled.
  2. Connect it the same way you connected your sandbox, so Claude can put the files it builds somewhere real.
  3. Tell Claude you want to build, not write. Say plainly what you want to make, for example: "I want to build a simple one-page website. Put the files in this folder." Claude creates the actual files and tells you how to open the result.
  4. Open what it made. Claude points you to a file (usually ending in .html). Open it, and you're looking at a real webpage you just made by describing it.
Mark complete when: Claude has created real files in your builds folder and you've opened the result.
Code 3Your first real build

Time to make something. Pick one thing from the menu, or bring your own, and ask Claude to build it. Start simple. You can always make it fancier once it exists.

Make something that runs

Describe it to Claude in plain language: what it's for, what should be on it, and any vibe you want ("make it warm and friendly," "use my brand colors, here they are"). Let Claude build it, then open it up and look.

Mark complete when: you've built one real page or tool and opened it.
Code 4Change it and watch it update

Building it once is good. Changing it is where you actually get powerful, because real things are never right the first time.

Go back to what you built and start asking for changes, the same way you'd give notes to a designer:

Make the change, open the page again, see the difference. Then do it again. This loop, describe, look, adjust, is the entire rhythm of building with AI. Once it feels natural, you're no longer a beginner. You're someone who builds.

Mark complete when: you've changed something you built at least three times and watched it update each time.

Code in the terminal

The full Claude Code experience, plus a real project to put it all together. This is the last section, and the one that opens the most doors. Do the first two sections first.

Terminal 1Why the terminal, and why it's less scary than it looks

The terminal is that plain text window you've seen over someone's shoulder and assumed was only for "real" programmers. Here's the secret: it's just another way to talk to your computer, and with Claude Code, you talk to it in normal English.

So why leave the friendly desktop app? Because the terminal is where Claude Code is most capable. It can work across many files at once, build bigger things, run them, fix them, and handle real projects, not just single pages. Everything you've done so far has been Claude with one hand tied behind its back. This unties it.

You're not becoming a programmer. You're learning to open one extra window and type plain requests into it. That's the entire barrier, and you're about to walk right through it.
Mark complete when: you believe, even a little, that the terminal isn't just for programmers.
Terminal 2Install Claude Code in the terminal

This is the one setup lesson of the section. Go slowly. If a step looks unfamiliar, that's normal, and Claude itself can help you through it.

  1. Open your terminal. On a Mac, it's an app called Terminal (search for it with Spotlight). On Windows, it's Terminal or PowerShell.
  2. Install Claude Code. Follow the official install steps at the Claude Code setup page. It's a copy-paste of one or two lines. Paste, press enter, let it run.
  3. Sign in. The first time you start it, it'll ask you to log in with your Claude account, the same paid account you've used the whole way.
  4. Point it at a folder. Navigate to your builds folder (Claude can tell you exactly how if you ask it in the desktop app first) and start Claude Code there.
If anything goes sideways: copy the error, paste it into the Claude desktop app, and ask "what does this mean and what do I do?" Using one Claude to help you set up the other is completely fair, and it's a habit worth keeping.
Mark complete when: Claude Code starts in your terminal and greets you.
Terminal 3Your first terminal session

It looks different, but it's the same conversation you've been having all along. Type what you want in plain English and press enter. Try a few small things to get the feel:

Notice that it can do everything the desktop app did, and more, because it's working right inside your folder with full reach. The rhythm hasn't changed one bit. Describe, look, adjust. You just have more power behind each request now.

Mark complete when: you've had a real back-and-forth in the terminal and built or changed something.
Terminal 4Pick a project, and where the trail goes next

You've got the whole toolkit now: co-work for thinking and writing, code on the desktop for making things, and the terminal for the real, bigger builds. The last move is to point it all at something that matters to you.

This is where a project finally earns its place. Not a practice exercise, a real thing you want to exist. Bring your own, or pick one:

Build something real

Work it the way you've learned: describe, look, adjust, repeat, until it's real and you're proud of it. Take your time. A project you care about is worth a few weeks of evenings.

When you're ready for people around you while you build: From Chat to Code is the solo path, on purpose. But building is better with a few others a step ahead and a step behind you. When you want that, AI Trailmates runs guided programs where you build something real alongside a small group. Come take a look
Mark complete when: you've started a real project of your own. There's no finish line here. There's just the next thing you build.

You did the whole thing.

Co-work, code, and the terminal, all of it. You've got the skills to keep building, so go make the next one.