In one week in October, 565 changes went into my code. 143 of them carry the signature of an AI co-author. I am the only person in the company.
I want to start with that number because most writing about AI and work is either a sales pitch or a eulogy, and the truth of it, for me, is neither. It is just how I work now.
Why alone
I don't have a team. I can't afford one. That is the honest reason, and I would rather say it plainly than dress it up as a philosophy.
What AI changed is not that I have help. It is that my ideas stopped being limited by my skills. Before, every idea came with a reading list. If I wanted to build something I had never built, I first had to learn the language, the framework, the edge cases, and by the time I had learned enough, the idea had usually gone cold.
Now I can go from an idea to a working version at a speed that still surprises me. Without AI I would not have built half of what I have, and nowhere near this quickly.
But speed is not the interesting part. The interesting part is what I still have to bring. I have enough base knowledge to know when the AI is building something properly and when it is building something that only looks right. That knowledge is the difference between moving fast and moving fast in the wrong direction.
So whose work is it?
Sort of mine.
The ideas are mine. The instructions are mine. And the part I have come to think matters most: finding the gaps. Noticing what is missing, what does not fit, what the system confidently did that nobody asked for. An agent will happily build the thing you described. It is much worse at noticing that you described the wrong thing.
If I had to say what authorship means now, it would be that: deciding what should exist, and noticing when what exists is not it.
What I would not hand over
When people ask what I would never let AI do, they expect a striking answer. Mine is: not much, really. I am overwhelmed most of the time. I run several products, the content for several accounts, and the systems that hold it all together. I want AI to take as much off my shoulders as it possibly can.
But I did draw one line, and I drew it into the systems themselves, not just into my head.
A proposal is not an approval.
My agents can draft, suggest, cut, render, schedule. They can propose a post, a reply, a change to something already queued. What they cannot do is publish in my name without me. That rule is written into the tools: a proposal goes into a store, and it stays a proposal until I say otherwise. The same goes for a validation check passing. Passing a check is not the same as being approved.
I did not do this because I distrust the tools. I did it because my name means something only if I stand behind what goes out under it.
The oldest lie on the internet
People have always lied online. "I made this much money last month, buy my course." "I have this many followers, pay for my thing and you will too." That last one is the easiest sell there is, because everyone wants more followers.
And it works, if followers are all you want. Growing a number is easy when the number is the point. The people who struggle are the ones trying to turn that attention into anything else, because attention built on a lie does not hold weight.
For my brand I always want the truth. I want to be trustworthy. Lying to get what I want is against my morals, because anything I built that way would be built on false pretences, and I would know.
AI makes that choice more important, not less. Every tool I use can now invent a moment, a quote, a result, a screenshot, in seconds and at no cost. So I wrote a second rule into my content system: never invent an event. My tools can retell something that really happened to me, in new words. If a moment is not in my real history, it did not happen, and it does not go on screen. Even when a made-up moment would perform better. Especially then.
Doom or utopia
People want me to pick a side on AI. Will it end in doom or create something like a utopia? Should we be worried?
I am very much in the middle. I build with it every day, and I can see both futures from where I sit.
What I have noticed is that the future is not decided only by the people training the biggest models. It is also decided, quietly, by the millions of us deciding what we let these systems do in our own small corners. What we let them publish. What we let them invent. Whether a proposal counts as an approval.
Those decisions are small. But they are the ones I can actually make. So I make them, and I write them down, and I build them into the tools so they hold even on the days I am too tired to check.
Top comments (0)