A few hours into my first session with Claude Code, I realized I couldn't stop. I wasn't tired. I wasn't forcing myself. I was just in it. That feeling of complete absorption. I hadn't experienced it with a computer in years.
By the end of the night, I had a fully functional personal website. Built from scratch. Me, someone who's always been intimidated by the command line, who viewed bash as a foreign language. A few hours earlier, I'd been debating whether to use Squarespace or WordPress.
I sat there thinking: what the fuck just happened?
I got my first computer when I was three years old. A 486 in the 90s Romania. I was lucky to have it. From the first moment, I was hooked. There was something about giving a machine a task and watching it respond. Something that felt like magic.
I'd spend days, weeks, playing games from CD-ROMs. Complete worlds contained on a disk.
Then I remember Yahoo Messenger. Suddenly I could connect to people I actually knew. My classmates. Send them emoticons, those goofy animated things that made sounds. The first time I used it, I knew: this is going to change everything. Maybe not the whole world, but definitely my world.
These moments are rare. You know them when they happen. Claude Code gave me that feeling again.
Here's what I understood after that first night: for my entire life, there have been three paths to building something.
Path one: buy a tool. Learn how to use it. Adapt your vision to its constraints. Pay for it forever. It's never really yours.
Path two: hire someone. Explain your vision. Wait. Get something back that's close but not quite right. Explain again. Wait again. Weeks of back-and-forth. What I call the context handoff problem. You're constantly translating what's in your head into instructions for someone else, and something always gets lost.
Path three: learn the craft yourself. Spend months, years, acquiring the skills. By the time you can build the thing, you've forgotten why you wanted to build it. Or the moment has passed. Or you're too busy building to think strategically about what you're building.
All three paths have the same underlying issue: friction between what you want and what you can actually make happen.
Claude Code eliminates the friction.
That first weekend, I didn't just build my personal website. I built my photography portfolio too. And the thing that made me understand this was different wasn't the output. It was the process. For the first time, I could iterate at the speed of thought. Change something, see it, change it again. No waiting. No explaining. No handoff.
It gave me wings.
What surprised me most was how Claude Code became my tutor.
When I didn't understand something, which was constantly, I just asked. What's an SDK? How does this work? Why did that break? And because it had my context, my codebase, my project in front of it, the answers actually made sense. They weren't generic explanations. They were specific to what I was building.
I learned more about how software actually works in a few weeks than I had in years of occasionally trying to pick up programming. The learning was a byproduct of doing.
A few weeks in, I connected Notion to Claude Code via MCP.
Our company Notion, like most startups, is a disaster. Documents that haven't been updated in years. Wikis that nobody maintains. We'd had a project to clean it up forever. Nobody ever did it.
I gave Claude Code a simple prompt: clean up this team space intelligently.
Five minutes later, it was done. It had analyzed every document, identified what was outdated, grouped related content, created logical folders, and trashed more than half of the material. Correctly identifying that it was no longer relevant.
Five minutes.
Think about what a human would need to do. Read every document. Understand the context. Decide what matters. Ask people if they still need things. Create a new structure. Migrate everything. That's days of work, minimum. Probably weeks.
It did make one mistake later. Trashed something by accident while trying to do something else.
This is when I stopped thinking about Claude Code as a coding tool and started thinking about it as something else entirely.
Steve Jobs had this line about computers being "bicycles for the mind." The idea was that humans aren't particularly efficient animals, but give us a bicycle, and suddenly we're more efficient than any creature on earth. Computers were supposed to be that bicycle for thinking.
For most of the history of personal computing, the bicycle has had training wheels, guardrails, and a limited number of paths you could take. You could only go where the software let you go.
Claude Code removes the constraints. You can ride anywhere.
What does anywhere mean? It means the terminal becomes the interface for everything. Writing documents. Building presentations. Analyzing data. Managing projects. Creating software. The boundaries between different types of work dissolve because everything flows through the same conversation.
This is hard to convey if you haven't experienced it. Software used to be a series of boxes. Each application was a box with its own rules, its own interface, its own learning curve. You moved between boxes constantly. Figma for design. Notion for documents. VS Code for code. Each one a separate world.
Claude Code collapses all of it into one place. The terminal. A conversation.
I've been thinking about what this means for software more broadly.
If I can build a website in a few hours, why would I pay for Squarespace? If I can clean up Notion in five minutes, why would I hire someone to do it? If I can iterate on designs directly in code, why do I need Figma?
The uncomfortable truth for a lot of software companies: if your product isn't deeply integrated into someone's workflow, if it's just a tool that does one thing, you're increasingly replaceable. Not by another product. By the user themselves, armed with Claude Code.
Canceling my ChatGPT subscription wouldn't change much. There's no lock-in. No deep integration. I use it from time to time, but it's not essential.
The products that will survive are the ones that go deep. The ones that accumulate context and integration over time. The ones where switching costs are real because the product knows things about your work that would take months to rebuild.
What I really want to say is simpler.
Every few years, something comes along that changes how you relate to machines.
The first personal computers. The internet. The smartphone. Each one expanded what was possible. Each one felt like magic when you first experienced it, and then quickly became normal, invisible, just the way things are.
Claude Code is one of those moments.
If you haven't tried it yet, really tried it, not just asked it a few questions, make yourself a cup of coffee. Clear a few hours. Start building something.
You'll understand.
You're reading this on the very website I built with Claude Code.