Command Line
The command line is a way of operating a computer where you type commands as text instead of clicking on icons. It's considered old-fashioned, but it remains the most important tool for programmers and in AI development to this day.
Most people operate a computer with the mouse. You see windows, icons, and buttons and click on what you want. But there is a second way to talk to a computer: you type a command as text and press Enter. The window in which this happens is called the command line. It usually shows only letters on a black background, plus a blinking bar waiting for input. The computer executes the typed command and also responds with text.
Why professionals prefer typing to clicking
At first glance this looks like a step backward to the 80s. In fact, the command line is significantly faster for many tasks. Anyone wanting to rename 500 files would spend hours clicking through menus with the mouse. With a single typed command, the same task is done in a second.
The second advantage is repeatability. You can write down a command, pass it on, and run it again as often as you like. You can’t do that with mouse clicks. That’s why instructions for developers almost always consist of commands to copy, not screenshots with arrows on them.
There’s also a practical reason: servers on the internet often don’t have a screen at all. A machine in a data center running a website or an AI model is controlled over the network. There’s nothing to click on there. Anyone who wants to operate such machines has to master the command line.
Structure of a command
A command usually consists of three parts. First comes the name of the program to be started. Then follow so-called options, which change the behavior of the program. Finally comes the target, i.e. the file or folder in question. One example is “copy -r pictures backup”: copy, including all subfolders, the folder “pictures” to “backup”.
The program that receives this input is called a shell. It is itself just a program, and it reads the line, breaks it down, and starts the desired tool. The shell can also chain several commands together. The output of one then becomes directly the input of the next, similar to an assembly line station passing the workpiece along.
It’s important to distinguish between terminal and shell, which are often confused. The terminal is just the window, i.e. the display. The shell is the logic behind it that understands what you mean. Common shells are called Bash, Zsh, or PowerShell. They differ in details, but the basic principle is the same everywhere.
The command line in the AI world
Anyone who wants to try out an AI model on their own computer today almost always ends up at the command line. Tools like Ollama or the Hugging Face libraries are installed and started with typed commands. Training a model is also triggered this way, often with dozens of options for learning rate, data volume, and compute time.
For some time now there has been an interesting feedback effect. AI assistants like Claude Code or Codex themselves run on the command line. You describe in plain English or German what should happen, and the program writes and executes the appropriate commands. This doesn’t abolish the command line, but instead gives it a conversational interface.
A typical misconception is the assumption that the command line is dangerous. What’s dangerous are individual commands that delete without asking for confirmation. That’s why you should never blindly copy commands from the internet, but understand them first. Otherwise, a terminal window is a perfectly normal program that you can close at any time.