I Thought It Was the Data. It's the Attention.

How can we effectively integrate AI into everything we do?

That is the line I wrote in June 2025, in a post called Maximizing AI Integration in Daily Life. My answer then was to track inputs (vision through a camera, hearing through a microphone, thinking through written notes) and feed all of it to an AI that would analyze my life back to me. I remember the shape of it. Lists. Four areas of human life to enhance. Next steps. It read like a plan because it was a plan, and I hadn't lived it yet.

A year later the shape is wrong in a way I couldn't see from inside it. Two moves, not one. And the move I thought was the point, feeding my data to AI, turned out to be the smaller half.

Capture stopped being something I do

The first thing that changed: I stopped feeding the AI.

Feeding your data to AI should be more seamless than that. You don't have to do much stuff — it's already gotten in your data. This is the new paradigm.

The crude version is what Meta is doing to its employees, tracking keystrokes, what they see on the screen, everything. The version I built on purpose is different: the vault, the agents that read what I'm already writing, the screen the agent can look at when I ask it what I actually did today. In February I called this the symbiotic experiment, and the word I used then was intimacy. It sees my time, it learns my voice. What I didn't say clearly enough is that the whole thing only works because capture became passive. In 2025 I was still the one doing the feeding. Now I'm not. The AI is with me constantly, not invoked.

Start with the computer. Extend it to the body later: eyes, ears, eventually. The frontier of integration isn't a better prompt. It's less distance between what I do and what the system sees.

I stopped being the guy doing the stuff

The second move is the one I'm still catching up to.

It's a change of mentality first. You stop being the guy who is doing the stuff and you become the manager.

There's a learning curve, and it isn't a software curve. I wrote this in Claude Made Me More Capable without fully hearing it. A steered fleet, not a single tool, the CEO sets direction, the agents run the work. I described the fleet model from the inside and still missed that the sentence was about me, not about the agents.

The scarce resource isn't data anymore. It isn't even hours.

Managing the agents — the attention is all you need. Attention is key.

I thought integration meant giving AI more of my life to analyze. It actually meant I do less, and steer more. The binding constraint on more work is how many agents I can steer well at once, not how many hours I have. Before I add a project I have to ask whether I have a free steering-slot, or whether I'm already steering all the agents I can hold. Spread too thin and they drift unwatched.

The learning curve is switching

The part nobody warned me about: switching projects has a learning curve. Switching work has a learning curve.

You have to be able to switch projects, switch work very easily.

The manager isn't the maker. The manager is the one who moves between the rooms. When I was the maker I could sink into one thing for a week (the song, the editor, the vault) and that was the old shape of focus. Now focus looks different. Now focus is holding six rooms in my head at once and giving each one the one good decision it needs from me this hour, then leaving. The agents do the rest. I said that too, in the earlier post:

I just need to guide them or steer them in the right direction in a few key important moments. That's it. The rest, they can take their own decisions, do their own thing...

I'm not good at this yet. I still catch myself dropping into a room and staying, doing the work, because doing is what I know. The shift from doing to steering is a real thing you have to learn, and I am learning it in public, and it is slower than I thought.

The frontier is the cockpit

Here is the part I don't have an answer for.

You need the tools to visualize what is happening. And now we are exploring this space — a lot of people are creating tools for that. We'll see what is the best way to manage a lot of agents.

I'm building one of them. TermCanvas, terminals on an infinite canvas, each agent in its own node. I'm building it because I need it, not because I think I already know the answer. There's a small space opening up right now, people are calling it the agent-cockpit space, or agent control planes, and nobody in it has a finished design. It's exploration, not solved.

How do you steer twenty agents at once and actually see which one is drifting. How do you know which room needs you next. I don't know. I don't think anyone does...

We are at that edge.