A plain-English explainer of the tool that builds most of what you see from me — including the videos that sent you here, and this page.
Ask a chatbot for help building something and you get a wall of code to copy, paste, and fix by hand. It cannot see your project, cannot run anything, and cannot check whether its own suggestion worked. You become the courier between the AI and the computer — and the courier job is most of the work.
Claude Code is Anthropic's AI agent that works where the work lives — your terminal, your code editor, or the browser. Point it at a project and it reads the whole thing, edits files across it, runs commands, and checks the results. Give it a bug and it fixes it, runs the tests, and looks at whether the fix actually worked.
The line that matters: it asks permission before it changes anything or runs anything. You stay in charge; it does the typing. (If you have seen my video "What an AI agent actually is" — this is that, for building software.)
This is the one tool on my list I can vouch for without qualification, because it built most of what I ship:
The pattern across all of it: I describe the outcome, it does the work in front of me, and I review what changed before it counts.
| Where | Terminal, VS Code and JetBrains editors, web, phone |
|---|---|
| What it can touch | Your files, your commands, your git — with permission, each time |
| What it costs | Needs a paid Claude plan (Pro, or Max for heavy use) — check current prices |
It can be confidently wrong — see my hallucination video. The people it burns are the ones who stop checking.
It earns its subscription when things ship. For questions and writing, the ordinary chatbot tier is plenty.
You do not need to be a programmer — I am a civil engineer — but you do need to be willing to look at what it did and say no sometimes.
The plans it needs are paid. If the budget is nothing, start with free chatbots and come back when a project justifies it.
Not an affiliate link. Anthropic does not run an affiliate programme, so I earn nothing if you click — this page exists because Claude Code built most of what I make, and people keep asking what it is. If that ever changes, this box will say so plainly.
One email a week: what actually happened in AI, what it means for you, and one thing you can use. Three minutes, no jargon, no hype. If a tool is bad I say so; if a week is boring the email is short.
Details checked August 2026 against Anthropic's own product page. Plans and prices change — confirm before paying for anything.