these 17 AI slop patterns are killing your content

Hey there, solopreneur!

I was scrolling through Twitter last week when I saw a thread from a founder I respect.

This guy built a $3M ARR SaaS, speaks at big conferences, has 50k followers.

But something felt off.

"This is a game-changer..." "Not because of X, but because of Y..." "Here’s the kicker."

The entire thread sounded like ai slop.

I've been obsessed with this lately. Over the past few months, I've spent hours analyzing content (both other people's and my own), trying to understand exactly what makes AI writing feel so robotic and LAME.

Because I actually love writing.

The process of getting my thoughts down, finding my voice, connecting with you through these newsletters. But AI has flooded everything with this same generic tone.

Your audience can smell it.

Once they suspect you're copy-pasting from an LLM, they're gone. The information is not wrong. But it's forgettable.

"Meh, that was decent" kills content creators.

So I've built a complete system to identify and eliminate AI slop patterns.

Today I'm sharing the exact playbook I use, my actual writing workflow, and how to set this up as a Claude skill that catches these patterns automatically.

why this matters now

When everyone sounds the same, a unique voice is your only moat.

Most people know AI writing feels off, but they can't name what's wrong. That's dangerous and you might be using these patterns without realizing it.

I'm about to break down the 12 most obvious AI slop patterns so you can spot them instantly.

(If you just want the shortcuts: scroll down for the complete anti-slop prompt you can copy-paste. But I recommend learning these patterns first, it'll help you spot them way faster).

the 12 patterns that scream "a robot wrote this" (prompt in the next section)

1. "No X. No Y. Just Z."

"No fluff. No theory. Just actionable insights."

AI loves this structure. You see it everywhere. Real humans don't talk in constant negation patterns.

2. "It's not about X, it's about Y"

"Success isn't about working hard, it's about working smart."

This binary contrast appears in every AI-written post. Sometimes useful, but AI overuses it to death.

3. Triple Threat Syndrome

"Fast, efficient, reliable." "Boost engagement, increase conversions, maximize ROI."

AI learned that grouping things in threes makes "good writing." So it does it constantly. Vary your rhythm. Two things. Four things. One thing. Mix it up.

4. "Game-changer" and "supercharge"

If you had a euro for every time AI wrote "game-changer," you'd buy OpenAI.

Describe actual change with specific metrics instead.

5. Overusing "real"

"Just real strategy from real experts getting real results."

Repeating "real" doesn't make you authentic. Specific examples do.

6. "To your success"

Email sign-offs like this instantly reveal AI wrote it. Even successful founders slip up and use this without reading their emails.

Sign off like a human.

7. Arrow overload

→ 10x your results → scale faster → join today

One arrow per post maximum. Maybe.

8. Emoji explosion

🚀 boost productivity now! 💡 revolutionary approach! ⭐ transform your business!

ChatGPT treats emojis like confetti. I don’t use emojis AT ALL.

9. Corporate verb disease

"...highlighting the benefits..." "...facilitating better outcomes..." "...leveraging synergies..."

Use simple verbs. "Show" not "highlighting." "Help" not "facilitating." "Use" not "leverage."

10. Thesaurus abuse

Unless your audience wears suits to Zoom calls, stop using:

• "utilize" (use)

• "implement" (start)

• "optimize" (improve)

• "execute" (do)

Write like you talk.

11. Hedging language

"It's worth considering..." "You might want to think about..." "It's important to note that..."

Just say the thing. Skip the diplomatic warm-up.

12. Short hook questions

"The best part?" "Want access?" "Ready to level up?"

Very 2023 ChatGPT energy. Ask real questions that need thought to answer.

13. "Enter: [thing]"

"Enter: my revolutionary framework"

Just introduce the thing normally.

14. "And here's the kicker"

AI's favorite transition phrase. Sounds like a 3am infomercial.

15. Em dash overload

"This approach—which many experts recommend—can transform your business—if you're willing to put in the work—and see real results."

Yes, writers use em dashes. No, not 6 times per paragraph. Mix your punctuation.

16. "X changed everything"

"This strategy changed everything."

Really? Every single thing? Use specific changes with actual numbers.

17. Symbol obsession

"This symbolizes..." "Which reflects..." "Emphasizing the importance of..."

Say what happened. Say what you learned. Skip the literary analysis.

Now you know how to catch all the slop phrases like Ash catches Pokemon.

train your claude skill (your automated slop detector)

Quick context:

Claude Skills let you give Claude permanent knowledge that applies to every conversation. Think of it like giving Claude a reference manual it always has access to.

In this case, you're giving Claude my complete anti-slop playbook. Now every time you write in Claude, it'll know these patterns and can flag them automatically.

Here's how to set it up:

Step 1: Enable Skills (if you haven't)

1. Go to your profile icon (bottom left in Claude)

2. Click Settings → Capabilities → Skills

3. Turn on "Skill Creator" (you need this to build custom skills)

Step 2: Create your Anti-Slop Skill

1. Start a new chat in Claude

2. Type: "Make a new skill"

3. Claude will ask you questions. Here's what to answer:

• What workflow should this skill handle? "Detect and flag AI slop patterns in my writing"

• Give concrete examples: "When I paste a draft and ask 'check this for AI slop', analyze it against the 20 patterns in the playbook"

• What makes this unique? "Use the Anti-AI Slop Playbook as the reference for all patterns"

4. Upload the playbook PDF when Claude asks for reference material: [copy this doc and download as pdf]

Step 3: Install the skill

1. Claude will generate your skill file - click Download

2. Go to Settings → Capabilities → Skills

3. Click "Upload skill"

4. Drag and drop the file you just downloaded

5. You'll see "Successfully uploaded" - your skill is now active

How to use it:

Paste your draft into any Claude conversation and ask: "Check this for AI slop"

Claude will flag patterns and suggest rewrites in cleaner language.

That’s all you need to do:).

Bonus: How I capture ideas naturally

One more thing that's helped me avoid AI voice:

I don't start by typing.

I use Wispr Flow to transcribe voice notes. Either on walks or at my desk. I pick a topic, ask myself a few questions, and basically podcast with myself.

The flow is way more natural than staring at a blank page. Filler words, tangents, half-formed thoughts - all fine. You extract the good stuff later.

Voice captures how you actually think. When you type, you self-edit before the idea hits the page. When you talk, the raw thought comes out first. We also tend use a more natural and authentic vocabulary.

It’s easier to write like you talk when you actually talk (lol).

Then I take that transcript and organize it.

But the come from my voice, not from prompting an LLM.

Try it. Your content will immediately sound more like you.

(Not affiliated with Wispr - it's just free and works well. Use whatever voice-to-text tool you prefer.)

Ole's Bookmarks

My YouTube Channel is live, you can check it out here 🙂 

ragebaiting is the most short-sighted marketing tactic out there. i’m happy people are starting to call it out

I’m diving deeper into speaking again and this channel is amazing. It takes an opposite stance of this overly performative way of talking. Highly recommend checking him out.

Speaking of workflows...

I've been running new experiments with AI content workflows for months now. Trying different approaches, testing what actually moves the needle.

One thing became clear:

the hybrid approach is what leads to long-term growth.

Human positioning/Deep Skill understanding + AI assistance without losing your voice.

That's the winning formula. But it requires a different skill set than what most people are teaching.

You need really specialized positioning. You need to spot bad AI writing in your own work. And you need to understand the DNA of authentic writing that actually convinces people.

I’ve been thinking about building something around this. Maybe a rebuilt version of the AI Audience Accelerator build around this hybrid approach.

Haven't made anything yet. Just want to see if people are interested.

If it sounds useful, click here to sign up for a waitlist (no commitment)

[Join it here]

Talk soon, Ole