When you use Claude Code, you are describing a piece of software you want to exist — and something on the other end is reading your description, deciding what files to create, writing the code, running it, noticing when it does not work, and fixing it, until the thing you described actually exists. The novel part is not that a computer writes code. It is that the input is English rather than a programming language.
The practical consequence: the question of whether a piece of software should exist is no longer bundled with the question of whether you personally can build it. Those two questions have been inseparable for sixty years. They are separable now.
The rest of this page — the interactive explainer below — walks through the five things Claude Code does with your description, with small animations you can click through.