The phrases that make LinkedIn comments sound like AI (and what to write instead)
A LinkedIn comment sounds like AI when it could sit under any post. Stock praise ("great insights", "thanks for sharing"), polished filler words ("delve", "leverage", "journey"), em dashes, hashtags and gushing are the usual giveaways. The fix is simple: point at one real detail from the post, add your own angle or question, and stop at two lines.
Why people spot AI comments so fast
Scroll the comments under any popular post and you'll see the same few lines over and over. They praise the post without showing that the writer read it. Readers learn the pattern quickly, and once they see it, they skip the comment and the name attached to it.
A comment works when it proves you read the post. That means mentioning something only that post contains: a number, a decision, a sentence that surprised you. Everything below is a version of the same mistake, saying something general when a specific line was available.
The phrases to drop, and what to write instead
These are the lines TwoCents refuses to write. Each one is fine in a real conversation, but in a comment box they read as filler.
| Phrase | Why it reads as AI | Write this instead |
|---|---|---|
| "Great insights" / "Valuable insights" | Praises the post without naming the insight | Name the one point you'd steal: "the bit about cutting the demo to five minutes" |
| "Thanks for sharing" / "Thank you for sharing" | The default sign-off of every generic comment | Say what you'll do with it: "sending this to our hiring lead" |
| "Couldn't agree more" | Agreement with nothing behind it | Agree with a reason: "same thing happened when we raised prices" |
| "This resonates" / "Really resonates with me" | Feels emotional but says nothing | Say which part, and why: "the Friday update part hit home, we tried it last year" |
| "Spot on" / "Absolutely spot on" | Short praise that fits any post | Add the detail that makes it right: "spot on about churn hiding in annual plans" |
| "Game changer" | Hype word, rarely true | Describe the actual change: "this would save our team a call a week" |
| "Food for thought" | Polite way of saying nothing | Ask the question it made you think of |
| "Key takeaway" / "Valuable lesson" | Sounds like a summary bot | Write the takeaway in your own words |
| "Well articulated" | Comments on the writing, not the idea | Respond to the idea itself |
| "Such an insightful post" / "Insightful perspective" | Generic praise in formal clothes | Quote the line you found insightful |
Plain praise isn't the problem. "Great post" followed by something specific is fine. "Great post!" on its own is not.
The words that give it away
Some single words show up far more in AI writing than in the way people talk. One of them won't sink a comment, but two or three in a short reply will.
- Big, vague verbs: delve, leverage, navigate, unlock, empower, foster, underscore, embark
- Corporate filler: landscape, synergy, paradigm, holistic, robust, seamless, ever-evolving, fast-paced
- Dramatic nouns: journey, tapestry, realm, testament to
- Inflated adjectives: pivotal, crucial, invaluable, remarkable, thought-provoking, insightful
- Stock openers: "In today's…", "Here's the thing", "At the end of the day", "It's not just…"
- Intensifiers: truly, absolutely, incredibly, 100%
If a word would sound odd said out loud to a colleague, it probably reads as AI in a comment too.
Habits that look machine-made
Words aren't the only tell. These patterns show up even when every word is ordinary.
- Em dashes. AI tools love them; most people typing a quick comment use a comma or a full stop. Break the sentence in two instead.
- Hashtags in comments. Hashtags belong on posts. In a reply they look automated.
- Bare praise. "Love this!", "Well said!", "So true!" with nothing after them.
- Gushing. Two superlatives in one reply ("amazing, truly incredible work") reads as fake enthusiasm.
- Correcting the author by accident. Lines like "the real question is…" or "that's the easy part" can sound like you're telling the author they missed the point, even when you agree.
- Emoji stacks. One emoji can be fine. Three in a row looks like a template.
- Length. A comment that runs past two lines starts to look like an essay. The best comments are one or two sentences.
Five rewrites
A founder shares that they hit their first 100 customers.
Before: "Huge congratulations! This is truly a testament to your hard work and dedication. Thanks for sharing your journey!"
After: "100 customers without a sales hire is the part I'd want to hear more about. was it all founder calls?"
Someone posts that they were promoted to team lead.
Before: "Congratulations on this well-deserved milestone! Wishing you continued success in this exciting new chapter."
After: "congrats Riya. first time leading people you used to sit next to is its own skill, hope the transition's been smooth"
A marketer writes that cutting their newsletter frequency raised open rates.
Before: "Great insights! This really resonates. Quality over quantity is a key takeaway for all marketers."
After: "weekly to fortnightly and opens went up is a nice counterintuitive result. did unsubscribes drop too?"
A recruiter posts tips for writing a better CV.
Before: "Absolutely spot on! These are valuable tips that every job seeker should leverage. Food for thought!"
After: "the point about putting numbers in the first line of each role is the one most people skip"
An engineer shares a postmortem of a production outage.
Before: "Thank you for sharing this thought-provoking post. It's a powerful reminder of the importance of robust systems."
After: "writing up the 2am rollback decision this honestly is rare. how long before you spotted the bad deploy?"
Every "after" does the same three things: it names a detail only that post contains, adds a thought or a question, and stays short.
A three-step check before you post
- Point at one thing. Can someone tell which post your comment is under? If not, add the detail.
- Add yourself. Your experience, your question, or what you'll do with it. That's what makes it a comment and not applause.
- Cut to two lines. Delete the opener, the sign-off and any word from the lists above.
How TwoCents uses this list
TwoCents is a Chrome extension that writes LinkedIn comments in your voice. Everything on this page is built into it: when you click Comment on a post, the reply it writes skips these phrases, words and habits, points at a real detail in the post, and stays within two lines. You can still change any word, make it shorter, warmer or bolder in one tap, and nothing posts until you press Post.