All lessons · 44
Setting up your AI environment: accounts, browser and five terminal commands
Two accounts, one browser, five terminal commands — a complete 15-minute setup for Windows, macOS, Linux and phone, with zero technical background and an honest list of things to skip.
Why you should care: you don’t need to “install AI”. You need a browser, two accounts and fifteen minutes. That covers every everyday task in this course. The second half of this lesson adds a black window with five commands — you only need it for the coding-agent lesson later in the track, and even then there are ways around it, as you’ll read below.
Five words before we start
- Browser — the program you use to open websites: Chrome, Safari, Firefox or Edge. If you are reading this page, you are using one right now.
- Account — your personal login for a website, usually an email address and a password. It is what keeps your chats available on any device.
- App / app store — an app is a program you install on your phone or laptop. Install only from the official store: App Store on iPhone, Google Play on Android, Microsoft Store on Windows. That one rule blocks most of the junk.
- Bookmark (and pinned tab) — a saved address in your browser, usually in the row under the address bar. A pinned tab stays open and jumps to the front. Both exist so you stop retyping addresses.
- 2FA (two-factor authentication) — a second step when logging in, usually a code on your phone. Even if somebody steals your password, they still can’t get in.
The 80/20: two accounts today, a third this week
The setup asks for two accounts today, not one. Why?
pick an answer — you get the explanation right away
| Account | Why | Cost |
|---|---|---|
| ChatGPT | Your daily workhorse | Free plan* |
| Gemini | Generous free plan, Google integration | Free plan* |
| Claude | The writing specialist, add this week | Free plan* |
* As of this writing, all three have a free plan, but plans and limits change every few months. Check the current offer on the official pricing page before you pay for anything — and don’t pay at all while you’re just learning.
Sign up with your existing Google account or email. Nothing else is needed today.
Check the current offer
Every “free” in this article means: there was a free plan when I wrote this. Companies change their plans often. Two minutes on the official pricing page tells you what is true today — and nothing in this course requires a paid plan.
The setup, on any device
Windows, macOS, Linux, iPhone, Android — the steps are the same everywhere:
Open both assistants in your browser
Go to chatgpt.com and gemini.google.com, sign in, and bookmark both (or pin them as tabs). Edge, Chrome, Firefox, Safari — any browser works; on a Mac, Chrome or Firefox make some AI sites behave a bit better. This is already a complete laptop setup.
Add the phone — this is where AI clicks
Install ChatGPT or Gemini from the official app store and only from there. If a website or an ad offers you an installer file, that’s a no. Then tap the voice button (a microphone or a sound wave) and ask something real: “I’m at the store, dinner ideas from what’s usually in my fridge”. Voice is the most natural interface ever shipped with AI — use it.
Optional: desktop apps
ChatGPT, Claude and Gemini all ship desktop apps, and the list changes often — check the official site. What they add: voice mode and one keyboard shortcut that summons the assistant over any program you’re in. Nice, not necessary.
Skip, for now: WSL (running Linux inside Windows — for developers), Python (a programming language), local models (AI running on your own machine, which crawls on a weak laptop), any “AI optimizer” utility, third-party “AI assistant” apps with big ads and subscriptions — and above all, any script a site asks you to run in PowerShell or Terminal that you don’t understand. A script is just a list of commands your computer obeys blindly. None of this helps a beginner; some of it is a great way to break your laptop.
What actually happens when you press Enter
Your message leaves the browser tab or the phone app and travels over the internet to the computers of the company that runs the assistant.
Click through all five and watch the green out› line under each one. Step 2
is the one to notice: what leaves your device is never your sentence alone —
it travels with the company’s instructions and the whole chat so far, which is
why the same question can get a different answer in a different chat.
Two practical consequences: your words are stored on someone else’s computer (so don’t paste secrets), and you need an internet connection (so offline plane mode won’t work).
One security step worth two minutes
You have two minutes for one security habit. Which one does the most for you?
pick an answer — you get the explanation right away
Protect the account, not the laptop
Your AI account will contain more about your life than your photo gallery. Turn on 2FA on your Google account — that single step protects ChatGPT, Gemini, Claude and everything else you log into with it.
And one habit: never paste passwords, card numbers or passport data into a chat. No real task needs it.
That’s the whole everyday setup. Everything from here to the end of the lesson is about one optional thing: the terminal.
The terminal, in five minutes
One black window stands between you and the coding agent later in this track. It’s called the terminal, and it has a reputation for being scary. It isn’t.
Not a terminal person? Read this first
This course goes through the terminal, because that is my own way of working and the most capable one: a model is a text-prediction machine, and the terminal is pure text — the two fit each other perfectly. For a developer, AI plus a terminal is the most powerful combination available right now.
It is also not the only way, and not mandatory. Almost the same things can be done through a normal interface: an editor like VS Code, or apps like Claude Desktop, Codex and Copilot, which do the work through windows and buttons instead of typed commands. And if you don’t want to code at all — keep using ChatGPT for everyday tasks. That works too.
I’ll cover all of these paths in the lessons ahead. We start with the terminal because once you have it, every other path is easier to pick up.
Three more words, so nothing below is mysterious:
- Terminal — the window. Also called console, shell or command line.
- Directory — just the old word for folder.
- Command — one line of text you type and send with Enter.
You can already open folders and run programs by clicking. What does typing the same thing as a line of text give you that clicking doesn't?
pick an answer — you get the explanation right away
Opening it — pick your system:
- Windows: press Start, type “Terminal” or “PowerShell”, press Enter. (On older Windows: “cmd”.)
- macOS: press Cmd+Space, type “Terminal”, press Enter.
- Linux: Ctrl+Alt+T, or find “Terminal” in the menu.
A window opens — black or white — with a line at the bottom ending in a blinking cursor. That line is the terminal asking “what next?”. Often it also shows where you currently are; that location is the thing to watch, because everything you type happens there.
The commands are identical on every system. Only the look of a location
changes: /Users/dana on a Mac, /home/dana on Linux, C:\Users\dana on
Windows. Read a path as an address, most specific part last.
1$ pwd2/home/dana3 4$ ls5Documents Downloads Music Photos recipes.txt6 7$ cd Documents8$ pwd9/home/dana/Documents10 11$ cd ..12$ mkdir my-first-page13$ cd my-first-page14$ pwd15/home/dana/my-first-pageFive commands, and that’s almost everything this track asks of you:
pwd— where am I?ls— what’s here? On Windows the word isdir(in PowerShell both work).cd Documents— walk into that folder.cd ..— walk one step back up. Two dots always mean “the folder above”.mkdir my-first-page— create a new folder. This is the one you’ll use to build a safe playground for your first agent.
Look at the second tab: ls is not a fixed answer, it’s a question about
where you are standing. Same command, two folders, two different answers.
That’s the one idea that makes the terminal click.
Three rules people trip over:
- Enter runs the command. Pasting just puts text on the line. Paste with Ctrl+V (Cmd+V on a Mac, Ctrl+Shift+V on most Linux terminals) or right-click.
- Ctrl+C is the way out if the terminal seems stuck — it cancels the current line or stops a running program. It deletes nothing.
- Spaces and case matter.
cd My Projectfails;cd "My Project"works. Easiest fix — no spaces in folder names.
Now try it yourself. Nothing here changes anything — it only looks around and makes one empty folder:
pwd
ls
mkdir terminal-practice
ls
cd terminal-practice
pwd
cd ..
pwdIf the second ls shows terminal-practice, you officially use a terminal.
When something doesn’t work
An install page confused you, or a command printed an error you don’t understand. Two habits cover both cases:
- For a command: read the last line of the error — that’s the actual
complaint. Then run
pwdandls: most errors turn out to be “you’re not where you thought”. If the terminal seems stuck, Ctrl+C gets you out. - For anything else: don’t guess. Describe what you see and let the assistant walk you through it, one sentence per message:
I'm setting up [which assistant / which command] on [Windows / macOS / Linux / iPhone / Android], and I'm stuck.
What I did: [the steps you took]
What I see now: [paste the exact wording of any message, or describe the screen]
I am not technical, so explain it in plain language.
Give me the next action as ONE sentence. After I do it, I'll tell you what happened, and you give me the next one.Step by step, one sentence per message. It works surprisingly well, and it beats closing the tab in frustration.
Do this today
- Create one account in your browser and bookmark the site.
- Turn on 2FA for the Google account you signed up with.
- Install one official app on your phone and ask it something out loud.
- Run the terminal practice sheet above — four minutes, nothing to break.
Then put the setup away and use it: pick one task from the everyday scenarios. And when you hit a limit and wonder what is actually free right now, Free AI tools has the honest answer.
The takeaway
- Two free accounts and a browser tab — that’s the whole everyday setup.
pwd,ls(ordir),cd,cd ..,mkdir— that’s the whole terminal.- Enter runs, Ctrl+C cancels, errors are text you can paste to an assistant.
- The terminal is the most capable path, not the only one — VS Code, Claude Desktop, Codex and Copilot do the same work through a normal interface.
What is the minimal complete setup for everyday AI on a laptop?
When you press Enter, where does the answer actually get written?
You want to know which folder the terminal is currently in. Which command?
A website offers a free "AI optimizer" for Windows that wants to run a script. What do you do?
Questions?
Something didn’t install or open? Describe what you see in the Telegram channel — someone there has hit the same wall.