Last week someone asked if I used AI on a post. The panic on their face said everything. I felt the shame spiral anyway.
Done right, it’s just a tool. This post is about how I make the donuts. It’s not about broad AI ethics or whether AGI will eat us. Just how I use it to say what I mean.
I don’t hide my AI use. But I watch people panic about being caught using it. As if exposure is the sin, not the work. The anxiety isn’t the tool. It’s about being seen.
Here’s My Actual Workflow
Let me walk you through what actually happens when I write a post. Say I want to write about why remote work fails in institutions that worship visibility. This stuff bugs me because I watch it play out in real time. Everyone dials in to prove they were there, even though the meeting made no sense.
Layer 1: Brainstorm & dig
I open a chat: “I want to write about why remote work fails in institutions that worship visibility. What angles am I missing?”
AI comes back with options. I’m not looking for answers. I’m looking for what pings something real in my thinking. Then reverse interview time. I set it up and the AI asks me what I actually think underneath the surface reaction. What’s the real problem? Why does that meeting still bother me? It forces me to stop pretending.
Layer 2: Finding the story
AI asks me for a scenario. I braindump a recent one: the meeting room was half-empty but everyone still dialed in to prove something. Nobody questioned it. The meeting itself was pointless. It could’ve been an email. But visibility mattered more than common sense.
Layer 3: Structure
AI brainstorms four or five angles on the story. I pick one, discard two, push back on the fourth. Back and forth. I’m not following it. I’m using it to find myself.
We figure out the shape. Hook with the pointless meeting. Build tension around why offices still demand physical presence when it serves nothing. Close on what actually matters when you strip away the visibility worship. Storytelling stuff.
Quick sanity check: Am I missing something obvious? Are there logical holes? This is where I fact-check what the AI suggested too. Not exhaustive, just the “did I accidentally write bullshit?” checkpoint.
Layer 4: Outline & draft
AI drafts an outline based on what we’ve figured out. I iterate until it feels like my thinking, not AI-speak. Usually two or three rounds.
Then either I write the post or AI does the shitty first draft. Depends on energy level. Today I’m tired, so AI gets to hold the pen. The result is… fine. It sounds like a “helpful blog guy”, not me. Smooth. Generic. Bloodless.
Layer 5: Voice pass
This is where it becomes mine. I tear through it. Compressed sentences. Word choice. Rhythm. Strip the smooth-sounding stuff. Take a chainsaw to the repetition. Make it sound like I’m talking.
For example: AI drafted “The paradox of remote work is that physical presence becomes performative.” I rewrote it: “Everyone dialed in from the office to prove something.” Same idea. Different blood pressure. One sounds like I’m thinking. One sounds like I’m explaining. I cut two pages down to one. I tighten every metaphor. I kill every qualifier that softens the argument.
This takes an hour. Or more. Worth every minute.
Layer 6: Polish
AI flags readability issues, flow problems, moments that land or don’t. I build a surgical editing plan. Fix this. Keep that (it’s intentional, readers need to sit in it). Ignore that suggestion.
I pick and choose. Some of its feedback is right. Some isn’t.
What This Actually Gets You
I used to spend three hours on structure alone. Now 30 minutes with the AI, then voice and story. The stuff only I can do. My posts are better.
I publish weekly now instead of waiting for inspiration to overcome friction.
The real mark: finished, clear, honest work that actually says what I meant.
What That Colleague Actually Asked
After I told them I used AI, they got quiet. Then: “Wait, so you still write the actual post?”
Yeah. I do the thinking, the outlining, the decisions, the voice pass. The whole thing that makes it mine.
“So you’re just using it like… a tool?”
Yeah. Like spellcheck.
Their whole face changed. The panic dropped. “Oh. Okay. I thought you just fed it a topic and published whatever came out.”
That’s the slop. That’s why people panic. Not because AI is shameful. Because people use it as shortcut and it shows.
No shame. No hiding. Just work.
