What Is Claude Code? The Tool That's Quietly Rewriting Software

The Short Version

Claude Code is an AI tool that writes code for you. You describe what you want in plain English, and it builds it. Not just a line or two. Entire features, bug fixes, full workflows. If you've read our article on vibe coding, Claude Code is one of the main tools making it happen.

It's made by Anthropic, the AI company behind Claude. And it's growing faster than almost any software product in history.

What It Actually Does

You know how you can ask ChatGPT a question in English and it writes you an answer? Claude Code works the same way, except instead of writing text, it writes the actual code that makes apps, websites, and software work. You describe what you want, and it builds it.

Think of it like having a junior developer who never sleeps, knows every programming language, and can read your entire project in seconds. That's Claude Code.

You open your terminal (the text-based interface developers use to talk to their computers) and type something like: "Fix the bug where users get logged out randomly." Claude Code reads the relevant files, finds the problem, writes the fix, runs tests to make sure nothing else broke, and saves the changes. You review what it did and approve it.

No typing code. No staring at error messages for hours. You describe the problem, it handles the rest.

Who Made It

Anthropic is an AI safety company founded by Dario and Daniela Amodei, both former executives at OpenAI. The company is focused on building AI that's powerful but also safe and controllable.

Claude Code was created by Boris Cherny, a staff engineer at Anthropic. It started as a simple research project in February 2025. A terminal tool that could chat about code and make small edits. Nothing revolutionary at first glance.

Then it got good. Really good.

Why It's a Big Deal

The numbers tell the story.

$1 billion in annual revenue in six months. Claude Code launched publicly in May 2025. By November, it hit a billion-dollar run rate. For context, ChatGPT (the fastest-growing consumer app ever at the time) took longer to reach that milestone.

4% of all public code on GitHub is now written by Claude Code. That's over 135,000 contributions per day, according to a SemiAnalysis report. They project that number will hit 20% by the end of 2026.

Companies are all in. Notion, Uber, Plaid, Zapier, and hundreds of others use it daily. Notion's co-founder said his job is now "keeping as many instances of Claude Code busy as possible."

How I Personally Use Claude Code

I use Claude Code every day. (I might be a bit addicted...)

This blog (the platform, not the article 👀) was built entirely by Claude Code. I did not write a single line of code.

When I was working at Meta, 99+% of my code was written by AI coding agents (Claude Code or similar). My job was just to manage a bunch of AI agents, like an orchestrator. I tell AI what to do, it does it, I review, done.

You might be wondering how Claude Code compares to tools like GitHub Copilot or Cursor. We'll break that down in a future article.

What This Means for Jobs

Here's the part nobody wants to talk about.

Tools like Claude Code are making individual engineers dramatically more productive. One person can now do the work that used to require a team of ten. I lived this at Meta. I wasn't writing code anymore. I was directing AI agents and reviewing their output. One person, multiple projects, shipping faster than a full team could a year ago.

The math is uncomfortable. If one engineer with AI can do the work of ten, companies don't need ten engineers. And they're acting on it. Layoffs across the tech industry have been relentless. Thousands of engineering roles have been cut at major companies, and the hiring that's coming back favors smaller teams working with AI, not large headcounts.

The role of "software engineer" isn't disappearing. But it's transforming. Less time typing code, more time thinking, reviewing, and directing. The engineers who thrive will be the ones who learn to work with these tools, not compete against them.

The Bigger Question

It might sound like anyone can be a software engineer now. Just type what you want and the AI builds it. And in some ways, that's true. The barrier to entry has never been lower.

But as a senior software engineer, I can tell you: a new skill has replaced the old one. It's no longer about being good at writing code. It's about being good at directing AI. Knowing what it's capable of and where it falls short. Knowing when to trust its output and when to push back. Knowing how to break a big problem into smaller pieces that AI can actually solve.

Think of it like managing people. Anyone can say "build me an app." But knowing how to ask, what to check, and what to watch out for? That takes experience. The skill has shifted from typing code to thinking clearly and communicating precisely.

The tool that writes 4% of the world's code today is only getting better. The question isn't whether AI will change software. It already has.

The question is whether you'll learn to work with it.