I Watched the Internet Arrive. AI Feels Exactly the Same.

Nobody can point to the day the internet arrived.

I lived through it. One year the web was a curiosity - a thing you read about, a screech from a modem, a handful of ugly pages. A few years later it was where your bank, your friends, your work, and most of your life quietly lived. There was no launch event. No line in the sand. It happened the way going broke happens in the old joke: gradually, then suddenly.

Here's the part I keep coming back to: almost nobody noticed it happening. People didn't sit around in 2001 saying we are living through the biggest change to human communication in a century. They just checked their email and got on with their day. The shift was invisible precisely because it was everywhere. And mostly, people didn't seem to care - the ground moved under them and they were looking at their shoes.

I'm watching the exact same thing happen again with AI. And once again, almost nobody is looking at the part that matters.

Most people are using a spaceship to check the bus schedule

Ask a normal person what AI is and they'll tell you it's a chatbot. You type a question, it gives you an answer. The slightly more advanced crowd - the "power users" - have worked out it can write things, so they use it to draft emails, summarise documents, plan a holiday. Genuinely useful. But it's a sliver of the story, and it's the sliver that looks most like things we already had.

We're using a spaceship to check the bus schedule and concluding that the spaceship is a slightly better bus schedule.

The interesting part of AI was never that it can answer. It's that it can act.

Because the thing that actually changes the world isn't AI that answers questions. It's AI that takes actions. The industry word is agentic, and all it means is this: instead of handing you a paragraph and stopping, the AI can go and do the multi-step thing. Open the files. Run the commands. Check the result. Fix its own mistake. Make the actual booking. Not "here's how you might plan the trip" - the trip, planned, with the reservations made. (I pull that one distinction apart in a shorter note.)

I'll give you a small, real example from my own week. I pointed an AI agent at a question and walked away. It searched a dozen different places for what people had actually said that month, weighed what it found, wrote the result to a file, noticed it had missed something, and went back for it - while I made coffee. I didn't operate a tool. I delegated a task and reviewed the output. That shift in posture, from operator to reviewer, is the whole thing in miniature.

That is a different category of object. An answer sits there and waits for you. An action goes off and changes something while you do something else. The first is a smarter encyclopedia. The second is a colleague.

Why this is a tectonic shift, not a feature

When software stops being a tool you operate and becomes something you delegate to, almost every assumption underneath our working lives bends. The unit of work stops being "the thing I typed" and becomes "the outcome I asked for." Even among the people building this, the understanding is thin. One engineer put it bluntly the other week: "very very few understand the rapid shift in AI architecture we're currently going through." He's right - and he's talking about the people who are supposed to be close to it.

I don't say this to sound breathless. I've been around long enough to be allergic to hype. I've sat through "this changes everything" about a dozen technologies that changed comparatively little. But I also lived through the one time the cliché was true, and the texture of this moment feels like that one - not like the hype cycles in between.

There's a generation coming that will never know the before. To them, software that carries out a goal on its own will be as unremarkable as a website is to a teenager today. They'll find it genuinely hard to believe the first generation of users looked at this and saw a better search box - the same way my own kids can't quite believe there was a time you simply couldn't look something up.

So I'm going to write it down. That's part of what this place is for: field notes from inside a paradigm shift, while it still looks like nothing much is happening. Future readers will have the benefit of knowing how the story went. We have the stranger privilege of being here before anyone's even sure it's a story.

I missed my chance to document the last one. I'm not going to miss this one.