Ideation.

Ideas come from conversations, podcasts, books, and the newsletters I read, most often from several of them colliding. Turning that collision into an idea is my job. It usually means finding the tension between two things that are both true, the connection nobody has drawn, or the reduction to what is actually true underneath the convention. When an idea is ready, I talk it through out loud, captured with speech-to-text, then argue it out with Claude until it is sharper than what I walked in with, or dead. What survives goes into an editorial backlog, where it can wait weeks for its turn.

Research.

Whatever I believe at the end of ideation goes into research as a hypothesis to test. The research phase, which I run with Claude Opus, gathers primary sources, numbers, and the strongest arguments against my position, and its job is to determine whether I am right. Sometimes the evidence carries a weaker version of the idea, or points somewhere more interesting. Then the idea changes. Every fact that reaches a published piece traces to a source I can point to. Claims that can't be verified get killed here. A lot of them do. I read the result before anything gets built on it.

Argument.

No prose gets written until the piece commits to three things: a thesis that could be proven wrong, a mechanism explaining why it holds, and one move a reader could make on Monday morning. The outline assigns evidence to each section, and where a detail from my own operating experience would land harder than any source, it asks me for that detail. The model is not allowed to invent one. I review the outline, answer its questions, and push back on the argument before a sentence of prose exists. Ideas that fail the test get reworked as ideas. Writing them more attractively is no fix at all.

Writing.

The draft runs against a voice specification I started building in May 2026 and have rewritten continuously since: how I sound and the constructions that give machine writing away. English is my third language. Working this way gets me closer to what I actually mean than writing unaided does. That is why I do it, and it took me a while to say so out loud. When a draft says something I do not mean, my comment sends it back to this stage.

Editing.

This is now the longest stage, and it runs as a loop. An in-depth edit works through the piece against the voice specification. Then a separate AI reader, one that has never seen my notes, my sources, or the draft's history, reads it cold, the way you would, and reports where it got lost. A targeted edit fixes what it found, or declines it with a written reason. Then the piece comes to me, line by line. I cut, rewrite, argue with it, and send it back, and the loop runs again. Most pieces take several rounds. Grammarly takes a final pass for spelling and grammar, which is the smallest job on this page. Everything here involves judgment (including the Grammarly recommendations I accept).

Finalization.

Once I sign off, the words are frozen. The last stage packages the piece without changing a word: it checks that every link resolves and that every source still says what the piece claims. Anything it finds, even a typo, goes back to editing. Nothing gets fixed in passing. Nothing ships until I approve the final package.

Time.

A 700-to-900-word article takes about four hours of my time, from the first spoken capture of the idea to the scheduled send. Those hours are spread over days or weeks. I don't count the thinking before that capture; it has no start time. I give you the number because it is what makes this different from a short prompt and a publish button. A monthly batch of LinkedIn posts takes an afternoon of my time to go from idea to being scheduled in Buffer.

Images.

I generate every image with ChatGPT Image or Gemini Nano Banana. Each one comes from a prompt written against a defined visual system, usually over several attempts. The hero image combines multiple outputs. I would rather make something specific to the argument than reach for another overused photograph from Unsplash or Pexels.

Audio.

Each article ships with a narrated version, generated with ElevenLabs from a customized voice that has nothing to do with mine. The player says "AI-generated," and the recording says so in its first sentence. If you ever hear my actual voice here, it will be because I recorded it.