The hands of a titan, the head of a human

Not long ago, on a train, I was reading Stephen Hawking’s A Brief History of Time. One passage struck me so hard that I wrote to my team about it right away.

Hawking describes the moment science outgrew the scientist. There was a time when one educated person could hold the whole picture of the world in their head. Then knowledge grew so vast and so technical that even scientists could only cover their own narrow patch. Science didn’t get worse. It simply got bigger than any one person could carry.

We are living through that same moment right now. Only not in science — in ordinary, everyday work.

We’re already there

AI agents have made each of us super-productive. A developer with Claude Code writes as much in a day as they used to write in a week. A product manager does a competitive analysis in a morning that used to take three days. A freelancer runs five clients the way they once barely managed two.

We’ve learned to do many times more. But we haven’t learned to remember many times more.

Research used to be slow and expensive, so you read every word you got back. Now you get thousands of pages in an hour. The problem is no longer how to produce them. It’s how to read them.

An ordinary day

At nine in the morning you open Claude, write a document, make three architecture decisions, create five files. At two you switch to another project, fix a bug, review someone else’s code. At five you analyse competitors. You close the laptop.

The next morning at stand-up: nothing. “What did I do yesterday?”

You definitely did it. You just can’t hold it all. The chat session ended, and the result dissolved into the history. Like a dream you know you had, but can no longer retell an hour later.

This isn’t a productivity problem. It’s a problem of scale. For the first time in history, we can do more in a day than we can recall by evening.

The paradox: the superpower created the gap

The very tools that gave us this power also created the gap. AI gave us the hands of a titan, but the head stayed human.

Chat is a great way to get something done and a poor way to keep it. The result lives in a tab for as long as the tab is open. Decisions stay behind in Slack, in transcripts, in conversations. Even when LLM memory becomes shared across vendors, it won’t work without a format that suits both people and agents.

Planners don’t save us either. They know what you planned, not what you did. Between intention and result there is a gap that memory used to fill. Now memory isn’t enough.

Saved chaos

The obvious answer is “put everything in one place.” But everyone says that, and people no longer believe it — about as much as they believe a promise to get married someday.

I tested it on myself. I asked two of my AI assistants to judge honestly whether xTiles had helped me become more organised. Both said much the same thing: the system taught me to collect better than to cut. Part of the chaos didn’t disappear — it simply became saved chaos. Duplicates, old versions, a backlog the AI generated that nobody moves.

One observation hit me in particular. Tasks with a date get closed. Tasks without a date or an owner don’t — even when an agent created them beautifully. Making a plan has become easy. Committing to it is as hard as it ever was.

Hence the rule I now keep: one active topic — one canonical project — one current state — one history of changes.

What we need is a centre of gravity

So what we need isn’t one more tool. We need a centre of gravity.

A place where the truth about your day flows in — automatically, without effort. Not what you planned. Not what you meant to do. What actually happened: the decisions you made, the documents you created, the tasks that appeared, and which AI tool each of them came from.

Imagine a day page that is empty in the morning and full by evening. Not because you filled it in, but because it collected for you. The next morning you see yesterday the way you could never reconstruct it from memory. Not a file log, not raw data, but a story that makes sense: here’s what you did, here’s what you decided, here’s what’s left.

The human is the one who decides

In a system like this, the human’s role changes. Doing a task, from now on, means first explaining what you’re going to do, why, and what result you got. Fewer and fewer tasks will need “hands-on” work in Figma or Amplitude — they’ll run through prompts. So every task becomes a conversation: with people or with agents.

And the climax is the moment a person reads a result written by several agents and makes a decision. Ticks a box or presses Approve. Writing matters because someone reads it. And more and more often, that someone will be us — a day, a week or a month after the work was done.

What changes

When the truth about the day collects itself, a team’s rituals change. Stand-up stops being an exercise in improvisation. The weekly report stops being fiction. The retrospective becomes a conversation about facts, not feelings.

AI has stretched what we can do beyond what we can remember. That power doesn’t need limiting. It needs what it lacks: memory, structure and an honest mirror of your day.

That is what we’re building at xTiles. Not one more place where you have to write things down, but a place where your work stays after you’ve closed the laptop.