The AI Writing Workflow I Actually Use (After Burning Out on "AI Will Do It All")

I had a phase where I thought AI would 10x my content output overnight. I'd write a one-line prompt, paste the result, publish it, and wonder why it felt hollow.

The posts that came out of that approach were technically fine. Readable, even. But they weren't mine. They didn't sound like me. And more importantly — they didn't connect with readers the way my slower, human-first writing did.

So I threw out the "AI does everything" approach and built something different. A workflow where AI is genuinely useful, not just a crutch.

Eighteen months later, this is the AI writing workflow I actually use.

Where Most People Go Wrong

Before the workflow, let me name the failure mode — because I lived it.

The "AI as ghostwriter" approach sounds logical: you give it a topic, it writes the post, you publish it. Speed. Scale. Output.

Except content isn't just information transfer. It's voice. It's trust. It's the specific way you see something that no one else sees quite the same way.

AI can replicate structure. It can generate paragraphs. It cannot replicate your experience running a business, raising kids, learning a hard lesson, or building something from nothing. Those are what make writing actually connect.

So the first principle of my workflow: AI handles the mechanics. You handle the meaning.

The Workflow

Step 1: I write the outline myself.

No AI here. Not yet.

I open a blank doc and think through what I'm trying to say. What is the one thing I want a reader to walk away knowing? What's the hook that makes them keep reading? What are the three points that prove my case?

I write this as rough bullet points. Not full sentences. Just the skeleton of the argument.

This takes 15-30 minutes. And it's the most important part. If the outline is weak, no amount of AI editing will save it.

Step 2: AI drafts the rough paragraphs.

Now I use AI — but narrowly.

I paste in my outline and give it a specific instruction: "Write a first draft based on this outline. Keep it conversational. Don't be precious."

The key word is first draft. AI is my first-draft generator, not my final writer.

What comes out is... raw. It's technically correct. It's also flat. It lacks the specific stories, examples, and personality that make content worth reading.

But it gives me something to react against. And that reaction is faster than starting from a blank page.

Step 3: I rewrite it with my voice.

This is where the real work happens.

I read the AI draft and ask: "Does this sound like me?" Usually it doesn't. The sentences are too formal, the transitions too smooth, the examples too generic.

So I rewrite.

I add a story from my actual experience. I replace generic advice ("consistency is key") with something specific ("I wrote three posts in a weekend and then nothing for three weeks — that's when I built this system"). I cut the parts that sound like every other blog post on the internet.

AI gives me clay. I shape it into something that could only come from me.

Step 4: AI checks for structural problems.

Once I have a draft I'm happy with, I use AI one more time — as an editor, not a writer.

I ask: "Does this post have a clear arc? Are there any logical gaps? Does the conclusion actually conclude?"

It catches things I miss when I've been staring at the same draft for an hour. Missing transitions. Points that don't support the main argument. Conclusions that trail off instead of land.

I take or leave every suggestion. The AI is a fresh set of eyes. Nothing more.

Step 5: I read it out loud before publishing.

This step costs nothing and catches almost everything.

I read the post out loud — actually out loud, not just in my head. Sentences that look fine on screen often feel clunky when spoken. Runs-on. Awkward transitions. Words I over-use.

If it sounds natural when spoken, it usually reads well too.

What AI Can't Do (And Why That's the Point)

The most valuable thing I can do for a piece of content is write from experience. AI wasn't there when I built my knowledge system. It didn't feel the frustration of blank page syndrome for months. It didn't learn, from scratch, how to turn notes into posts.

That's mine. And it's what makes content actually useful.

AI is good at: generating first drafts, catching structural problems, suggesting alternatives when I'm stuck on a sentence.

AI is bad at: knowing what only I know, understanding what my specific reader needs to hear, creating the emotional connection that makes someone share a post with a friend.

The workflow respects that distinction. AI on the mechanics. Human on the meaning.

Why This Is Faster, Not Slower

You might be thinking: "That sounds like more work than just writing it myself."

Sometimes it is. For short pieces or ideas I'm very clear about, I write from scratch and skip most of the AI steps.

But for harder topics — pieces where I have rough ideas but haven't fully thought them through — this workflow is genuinely faster. The AI first draft breaks the ice. The back-and-forth between my voice and its structure gets me to a finished post in 60-90 minutes instead of three hours of staring at a blank page.

The workflow scales with the difficulty of the topic. Simple ideas, I write myself. Complex ideas, AI helps me start.

The One Rule I Don't Break

AI assists. I decide.

Every word in a published post is either written by me or rewritten by me. I never publish AI output without making it mine first.

This isn't a moral stance. It's a practical one. Posts that sound like me connect with readers. Posts that sound like AI — even good AI — disappear into the noise.

Your voice is the one thing AI genuinely cannot replicate. Protect it. Build workflows that amplify it, not ones that replace it.