Hey Reader,

Look, today's newsletter isn't about whether we should or shouldn't be using AI in our writing.

I have thoughts on that. [You can read them here.] I'm on the side where AI isn't going anywhere, and I absolutely love using it as a creative thought partner. I think it's great for creativity and your thought sparring process, but I also understand the arguments against it.

That debate? We could have it all day. That's not what this is about.

Today's newsletter assumes AI is here to stay and most of us are going to use it and deal with it at some point.

So, my focus and angle are on helping you make AI content yours.

This is about being involved in every step—before, during, and after—and having access to go back and forth with the AI in the original context.

If you don't have that? I'm sorry you're working with or for people who are giving you shit to polish, and this isn't going to help you with that.

Coolies?

Cool.

My process (the real behind-the-scenes)

Before I show you how to edit AI content, let me tell you how I actually use it.

Because I'm not proclaiming this is THE method, it's simply MY method. You may have a different method/order of operations that works for you, and that's fantastic.

My mindset going into AI advice (or really any advice):

Don't: Try to fix what ain't broken

Do: Constantly learn from others so you can tweak as needed

Step 1: Set up for success (you're the leader, not the follower)

I have a Claude project specifically for my newsletter. It's got every single newsletter I've ever written uploaded, so it understands my voice, tone, and style.

I also have custom instructions that tell it things like:

• Target length (around 1,200 words)
• Never make anything up. If you have contextual questions, ask me.
• Avoid phrases AI overuses (e.g. "Here's the thing:")
• Write conversationally, like you're talking to a friend
• Use my frameworks and examples when relevant

But here's the most important part: I NEVER start with a blank page.

This is crucial because it establishes who's leading the conversation from the very first prompt.

I always begin with something I've already written. For example, when I started this newsletter, I didn't just say "write me a newsletter about AI editing." Instead, I said:

"I want to write a newsletter about how to edit AI content. Here's what I'm thinking: [then I pasted several paragraphs of my thoughts about the state of AI writing, my process, and what I wanted to cover]."

Or when I was working on my recent newsletter about content frameworks, I started with: "I've been thinking about how content pillars are holding people back. Here's my take: [followed by my actual thoughts in my own words]."

This approach does two critical things:

1. It gives AI something good to work with. Claude's own best practices suggest starting with content in your style/tone produces way better output than asking it to write from scratch.
2. It establishes that you're driving this thing, not the other way around. You're not asking AI to think for you. You're asking it to help you organize and expand on thinking you've already done.

This mindset shift—that you're the leader, not the follower—is everything.

YOU ARE THE BIG DUCK, NOT THE LITTLE BABY DUCKS 🦆

Step 2: Developmental editing (the back-and-forth)

This is honestly the most important part of the entire process, and it's where most people stop way too early.

I always ask for an outline first. Always. Even for shorter pieces.

For this newsletter, my first prompt after giving my initial thoughts was: "Can you give me an outline for this newsletter based on what I've shared?"

Then I look at the structure and ask myself: Does this flow make sense? Is this the right order? Are we missing anything important? Are we including anything that doesn't belong?

Then comes the back-and-forth:

"This part doesn't work because it jumps too quickly from setup to tactics without explaining why the setup matters."

"I'm thinking we should move the caveat about polishing turds earlier because people need to know upfront if this applies to them."

"This section on prompts feels too long. The whole point is that prompts aren't everything. Actually, on second thought let's remove it. I don't want to get into a fight with prompt engineers about prompts that detracts from my bigger point."

"Add more about the mindset shift—that you're leading, not following."

Here's what I'm doing in this phase: I'm using my editorial judgment to shape the content before we even get to the writing. I know what should be included and what shouldn't be. I can feel when the structure is off.

This isn't something I can teach you in a single newsletter (I have a whole course on developmental + copy editing if you want to dive deeper), but you have to be able to recognize:

• When arguments are in the wrong order
• When you're missing critical context
• When you're including stuff that doesn't serve the main point
• When the flow doesn't feel natural

Again: You're the leader. You have to know what you want this piece to accomplish and be able to guide the AI toward that goal.

Only after I'm happy with the structure do I say, "Okay, now let's write it."

Even then, I'm not done with developmental work. As it writes, I'll interrupt:

"Stop. This paragraph is trying to do too much. Break it into two separate ideas."

"This example doesn't work. Can you use something more specific?"

"This transition is clunky. We need a smoother bridge between these sections."

The key is that you're staying involved throughout the entire creation process, not just editing at the end.

Step 3: Apply the HUMAN method

Once the structure is solid, I focus on humanizing it.

This is where most people think AI editing begins. But it's actually the final step.

The HUMAN Method

H - Humanize the Voice

If you don't set AI up for success—if it doesn't know you, doesn't understand what you're trying to do—then yeah, it's going to write like it went to business school and never recovered.

But even when it does have your context and style, it will still default to that formal tone if you don't catch it.

For example, even my well-trained Claude newsletter project might write: "This approach has proven effective in multiple client scenarios."

I'll change it to: "This approach works like gangbusters (I've tested it with 47 clients and counting)."

The difference? The second version sounds like me talking to you, not like me presenting to a board of directors.

Look for:

• Formal language that needs loosening up
• Places to add verbal tics, parentheticals, personality quirks
• Opportunities to sound like you're actually talking to someone
• Sentences that sound like they came from a corporate memo

U - Unpack the Generic

AI loves vague, safe statements. Your job is to make them specific.

When I see something like "This strategy delivers results," I ask my favorite editing question: "What do I mean by this?"

Then I'll change it to: "This strategy helped Sarah go from 12 Instagram followers to 2,400 in three months, which led to her first $5K client."

Specific examples do three things that generic statements can't:

1. They prove you're not just theorizing. Anyone can say, "This works." Specific results show you've actually done the work.
2. They help readers visualize success. "Get results" is abstract. "Go from 12 followers to 2,400 in three months" is something they can picture happening to them.
3. They stand out in a sea of sameness dolphins (I dunno, I'm so over that phrase, but it also anchors my point, but I can't leave it, so dolphins). When everyone's saying "this approach is effective," you're the one showing exactly how effective.

Always look for:

• General claims that need specific examples
• Vague benefits that could be quantified
• Abstract concepts that need concrete illustrations
• Numbers that could be more precise (not "many clients" but "23 clients")

M - Make it Personal

AI can't share your experiences. Only you can do that.

This is soooo key because our experiences are by far the easiest way to stand out (always), but especially right now.

When everyone's writing stuff with AI—especially when people aren't doing the work of actually giving a shit and being involved in the before, during, and after—it's going to produce whatever has already been produced on the internet.

In other words, it's regurgitating a garbage dump.

Your experiences are unique. No one else has been through exactly what you've been through.

AI might write: "Many professionals struggle with time management."

I'll change it to: "Last month, I talked to three different clients who all said the same thing: 'I'm drowning in my to-do list.' One was working until 11 PM every night. Another hadn't taken a real lunch break in six weeks."

AI doesn't know your stories. It doesn't know what you went through last week. It doesn't know about the client call that made you rethink everything, or the moment you realized you'd been doing something wrong for years.

For the personal stuff, you might need a different approach. Maybe a different project where you upload your call transcripts and ask AI to pull out themes, moments, quotes.

Whether you're doing this for yourself or for a client, use AI to help you identify those golden moments—then choose which ones to weave into your content.

Add:

• Your client stories (with permission, obviously)
• Your personal experiences and realizations
• Real examples from your work
• Moments that changed your perspective
• Failures that taught you something important

A - Add Your Perspective

AI is annoyingly balanced. It doesn't want to offend anyone.

But you have opinions. USE THEM.

AI is spreading uninsightful insights everywhere, which means sharing your unique insights is a layup. (The easiest two points you can get if you aren't a basketball person.)

What is your point of view? What is your perspective? That's why people are reading you and not just anyone.

AI: "There are various approaches to this challenge, each with merit."

Me: "Most productivity advice is bullshit because it assumes you have an assistant and no kids." (FOR REAL THOUGH. GET OUT OF HERE YOU 4AM BROS. SHOO!)

If you're one of the many people spreading uninsightful insights all over the internet, this is your chance to show your ideal audience what you think about the status quo. What's your opinion on the problem? What do other people get wrong about it?

Look for opportunities to:

• Challenge conventional wisdom
• Share your contrarian takes
• Stop hedging and take a stand
• Show what you think others get wrong
• Explain why the status quo isn't working

N - Nail the Emotional Impact

The best writing makes you feel something. AI writing makes you feel...nothing.

This is where you separate yourself from the flood of robotic content that's technically correct but emotionally flat.

After editing, ask yourself: "Will my reader feel something when they read this?"

If not, find places to add:

• Surprise ("This completely changed my mind about...")
• Validation ("I know exactly how frustrating this is...")
• Hope ("Here's what's actually possible...")
• Curiosity ("Wait until you see what happened next...")
• Relief ("You're not crazy for thinking this...")
• Urgency ("The cost of waiting is...")

Neutrality doesn't move people to action. Neutral doesn't make them think differently. Neutral doesn't make them remember what you said three days later.

Your job is to take that neutral information and infuse it with the emotional weight it deserves. Make them care. Make them feel something. Make them unable to scroll past without engaging.

The difference between content that gets ignored and content that gets shared often comes down to emotional impact.

Common mistakes (that make me twitch)

Mistake #1: Skipping developmental editing. You can't polish a turd. If the structure is wrong, no amount of voice editing will fix it.

Mistake #2: Accepting AI's formal tone. AI defaults to "professional." Your audience wants "human."

Mistake #3: Not starting with your own thoughts. Garbage in, garbage out. Give AI something good to work with.

Mistake #4: Treating it like proofreading .Real AI editing is about making it sound like you, not just fixing typos.

So...

Stop treating AI like a ghostwriter you've never met. Start treating it like a thought sparring partner who needs your brain, experience, and personality to make the content actually worth reading.

And hey, if you're wondering...this newsletter was AI-assisted but heavily edited by me for about 2 hours. The thinking is mine. The experiences are mine. The arguments are mine. The voice is definitely mine.

The AI just helped me organize my rambling thoughts into something you'd actually want to read.

Which, honestly, is exactly what good editing has always been about.

Cheers,

Erica

P.S. If you want to get better at editing (AI or otherwise), my Content Editing 101 course teaches you exactly how to turn decent writing into great writing. It's got frameworks, examples, and AI bots that help you apply what you learn. Check it out here.

PPS. My latest podcast on my other newsletter, How Solos Scale, just dropped. Go check it out and subscribe if you're a solopreneur keen to scale past $20k a month without working more hours.​

​

New here? I'm Erica. Your seltzer-loving solopreneur who helps you earn more money with content that moves people to action. If you loved this email, but aren't on the list, subscribe here. And say hi on LinkedIn & Substack, or check out my website for more. ​​​