The 6-Step Editing Workflow

Module 4 · Section 2 of 5

The 6-Step Editing Workflow

Each piece you write with AI should follow this sequence. The whole workflow runs 85-120 minutes for a typical 800-1500 word piece.

Step 1: Start With Your Idea and Outline (10-15 min)

AI can’t do this part. This is where your thinking happens.

This takes 10-15 minutes but it’s the most important step. If you skip this, AI will give you generic structure based on what it thinks you want to say — not what you actually want to say.

What this looks like: your core argument in 2-3 sentences, main points as 3-5 bullets, personal insights or examples only you can provide, questions you want to answer, and what makes this different from other content on this topic.

Without your outline, AI defaults to template thinking. It produces the structure everyone else uses for this type of content. With your outline, AI helps you execute your vision — the specific argument you want to make in the specific way you want to make it.

This is the difference between AI as a writing partner and AI as a writing replacement. The partner executes your thinking. The replacement gives you everyone else’s thinking.

Step 2: Use AI for Research and Structure (15-20 min)

This is where AI shines. Let it gather information, suggest organization, and find connections between ideas.

Effective prompts look like this:

• “Here’s my outline. Help me research X, Y, and Z. Suggest how to structure this for maximum clarity.”
• “Find recent examples of [topic].”
• “What are the counterarguments to [position]?”
• “Suggest transitions between these sections.”
• “What context am I missing that readers need?”

AI is excellent at this kind of work. Let it do the heavy lifting on research and logical flow while you focus on the aspects that require your judgment.

Common mistakes here: asking AI to “make it better” without specifying what better means, accepting research without verifying sources, letting AI decide structure instead of suggesting structure for your approval.

AI should support your decisions, not make them for you.

Step 3: Write Bare Bones First (15-20 min)

Put in your basic argument, key points, and personal insights — the stuff only you can say. Writing 200-300 words of core content yourself before letting AI expand anything anchors the piece in your actual thinking.

What to write yourself: the opening hook (your unique angle), core thesis statement, personal stories or examples, controversial or distinctive opinions, concluding insight.

These are the parts where your voice matters most. Don’t outsource them. If you let AI write your opening, you’re letting AI decide what matters. If you let AI write your conclusion, you’re letting AI decide what the takeaway should be.

Those decisions should be yours.

Step 4: Let AI Fill Gaps (15-20 min)

But only where you decide gaps exist. Not where AI thinks they are.

Good gap-filling requests are specific:

• “I need a transition between section 2 and 3.”
• “I want to expand on this point about X with supporting evidence.”
• “I need an example that illustrates Y.”
• “Explain the technical background readers need to understand my point.”

Bad gap-filling sounds like: “Make this better.” “Add more content.” “Expand this section.”

What do these requests actually mean? Better how? More content about what? Expand in which direction? AI will guess, and its guess will be generic because it doesn’t know what you’re trying to accomplish.

Common mistakes here: asking AI to expand sections that don’t need expanding, accepting AI’s suggestions for new sections without evaluating if they serve your argument, letting AI fill gaps you didn’t know existed until AI told you they existed.

Your judgment determines what gets added. AI is the research assistant, not the editor.

Step 5: Edit Ruthlessly (20-30 min)

This is where voice comes back.

Cut the AI-isms. Add your personality. Be honest about what you actually think. Remove anything that sounds like it could have been written by anyone.

Cutting 20-30% of what AI generates and rewriting another 20-30% in your own words isn’t a bug — it’s what turns “technically correct but soulless” into “technically correct and distinctively mine.”

What to cut: generic superlatives without specific claims (“game-changing,” “revolutionary”), hedging language that softens everything (“it might be argued that”), unnecessary qualification that adds no value (“It’s worth noting that”), corporate platitudes that mean nothing (“leverage synergies”), anything that sounds like a press release instead of a person.

What to add: your actual opinion (not the diplomatic version AI produces), specific examples from your experience (not generic examples from AI’s training data), honest acknowledgment of limitations (not fake confidence in uncertain claims), your characteristic phrases (the stuff your style guide might have missed), conversational asides that show how you think.

The key question to ask yourself: would someone who knows you recognize this as yours before seeing your name?

If no, keep editing.

Step 6: Read Everything Aloud (10-15 min)

If it sounds like a corporate presentation instead of you talking, rewrite it.

This catches more problems than any other technique. Reading aloud significantly improves error detection compared to silent reading — and people often don’t realize they perform better when reading aloud, which is why this step gets skipped so often.

What you’re listening for: rhythm and flow (does it move naturally?), natural breath points (can you say these sentences without gasping?), words you would actually say (or words you’d never use?), sentences too long or complex (does your voice sound like this?), transitions that feel forced, places where you lose energy.

Half of editing issues show up only when you read aloud. No exceptions to this rule.

The read-aloud test reveals what visual editing misses. Text can look fine on screen but sound robotic when spoken. Trust your ear.