## The Excuse Expired
I scrolled past four different B2B posts last week doing the exact same thing: an em-dash standing in for a period, a list padded or trimmed until it landed on exactly three items, a hook manufacturing vulnerability out of thin air with no fact behind it I could check.
A year ago, I would have shrugged. Everyone was still working out what AI-assisted writing should sound like, and a few rough tells were the price of moving fast.
That excuse is gone. Every tell below takes about two minutes to catch, using tools sitting in the same tab that generated the slop in the first place. Use AI to draft all you want. Just stop shipping the first pass. Leaving these tells in signals one thing: checking wasn’t worth your time.
Six Tells Still Showing Up in B2B Content
These show up everywhere: LinkedIn posts, website copy, cold outreach, case studies. Not just LinkedIn.
1. The em-dash doing a period’s job
**Slop:** “AI drafts our content — then edits itself — before a human ever sees it.”
**Fixed:** “AI drafts our content. Then it edits itself. A human sees the final draft.”
2. The “it’s not X, it’s Y” reveal
**Slop:** “This isn’t a content problem. It’s a leadership problem.”
**Fixed:** “Leadership created this content problem.”
3. Generic superlatives with nothing specific behind them
**Slop:** “In today’s fast-paced digital landscape, AI is revolutionizing how B2B teams unlock growth.”
**Fixed:** “AI cut our first-draft time from three hours to twenty minutes. It also made every post sound identical.”
4. The list forced to exactly three items
**Slop:** “To win in 2026, you need speed, personalization, and authenticity.”
**Fixed:** “To win in 2026, you need speed and personalization. Authenticity is table stakes, not a differentiator.”
5. The fake vulnerability hook
**Slop:** “I failed. I cried. Then I learned the one lesson that changed everything.”
**Fixed:** “I lost our biggest client in Q2 because I sent the proposal to the wrong contact. Here’s the check I added to every send since.”
6. Hedge words stacked on top of each other
**Slop:** “To some extent, arguably, this could potentially help improve engagement in certain contexts.”
**Fixed:** “This increased engagement 12 percent in our test. It won’t work for every audience.”
The Fix, Part One: Edit Like Someone’s Watching
Before you hit publish, run this pass:
- Find every em-dash. Read the sentence with a period instead. If the meaning holds, use the period.
- Find every “it’s not X, it’s Y” sentence. Cut the setup and state the real claim.
- Find every superlative with no number or name behind it (game-changing, unlock, revolutionize, fast-paced). Attach a specific result or delete the word.
- Count your lists. If one landed on exactly three items, check whether three is true or just convenient. Let it be however many items are real.
- Find your opening hook. If it names a failure with no detail someone could fact-check, name the real one or cut the hook.
- Find stacked hedge words (to some extent, arguably, in many ways, potentially). Pick one claim and commit to it.
Then run the read-aloud test. Read the paragraph out loud, to an empty room if you have to. If you would not say that sentence to a client’s face in a meeting, delete it and write the sentence you’d actually say.
The scalable version
Ruben Hassid’s approach turns this checklist into a standing rule instead of a one-time pass. Build a file (he calls it anti-ai-writing-style.md), or grab his free pre-made one at [how-to-ai.guide](https://how-to-ai.guide) (it asks for your email, skip the paid upsell), listing every word, phrase, and pattern you’re banning: delve, unlock, leverage, “let’s dive in,” “it’s not X, it’s Y,” em dashes, forced rule-of-three lists. Upload it to whatever AI tool you draft with, and open every session with: “Read my anti-AI writing style file first. It contains every known pattern of AI writing I want to avoid. Apply these as rules to everything you write for me.” Then ask it to audit each draft against that file before you publish.
The file does the work. The prompt just points at it.
The Fix, Part Two: Feed It What It Doesn’t Have
AI has no opinions of its own and no lived experience to draw on. It only has patterns pulled from everything already written, which is exactly why so much AI-assisted content sounds the same. It’s remixing the same source material everyone else is remixing.
The fix is to feed it the one thing it’s missing before it drafts a single sentence: your actual opinion, in your actual words. That’s easier to do by talking than typing. Dictate your real take on a topic, unfiltered, then hand the transcript to AI and ask it to structure and tighten what you actually said, not invent something new.
I’ve been using Typeless, a voice keyboard that makes you smarter, to do exactly this. Use my link to [join and get a $5 credit for Typeless Pro](https://www.typeless.com/refer?code=IODFHB6).
Full transparency: that’s a referral link. I get a small kickback if you sign up.
Before You Hit Publish
Run these five questions on anything you’re about to post:
- Did I say something here that only I could have said, based on something I actually did or saw?
- If I strip out every superlative, does a real claim survive?
- Would I say this sentence out loud to a client’s face?
- Is my list this length because it’s true, or because three sounds clean?
- Am I hedging because I’m genuinely unsure, or because committing feels riskier than qualifying?
If you can’t answer all five, keep editing. AI wrote the draft. You still own what goes out under your name.