JDKola reflects on his process of using AI tools like Claude and NotebookLM to draft and develop article posts — not as a shortcut, but as a deliberate practice of protecting and amplifying his narrative identity. A builder’s honest account of working with AI without losing himself in it.
The tool doesn't write for me
There’s a version of this story where I say AI does the heavy lifting and I just clean it up. That would be easier to say. It would also be dishonest.
What actually happens is messier. I start with a concept — sometimes a clear idea, sometimes just a feeling I can’t fully name yet. I write it out. Long, unstructured, sometimes going in circles. I spend hours on that part alone. That part is mine.
Then I bring it to Claude or ChatGPT. Not to generate the post. To pressure-test whether what I’m trying to say actually holds together when someone else — something else — tries to shape it.
What comes back is a draft. And that’s when the real work starts.
I read it looking for words that don’t sound like me. Phrases that are technically correct but emotionally wrong. Sentences that are polished in a way I’m not. I edit those out. Not because the AI failed — but because my job is to notice the difference between what sounds good and what is actually mine.
The AI fortifies my identity. It doesn’t construct it.
That distinction matters more than I can fully explain right now, but I know it’s true because I feel the difference when I read something back and it sounds like me — and when it doesn’t. The feeling is specific. It’s not about quality. It’s about alignment.
After posting, I don’t just move on. I sit with it for a few days. Let the emotion of writing it clear out. Then I come back and read it like a stranger would. That’s usually when I catch anything that doesn’t sit right — a structure that drifts, a line that performed sincerity instead of expressing it, a moment where the narrative identity slipped.
I revise then. Quietly, without announcing it. Because the work isn’t done when it’s published. It’s done when it finally represents what I meant.
That’s what I’m building toward. Not just better posts. A consistent voice that I can recognise and defend — one that AI helps me hear more clearly, not one it invents on my behalf.
I’m still learning how to hold that line. But at least I know the line is there.




