Free check
Does your LinkedIn post sound like AI wrote it?
Paste a post. You get a verdict and every tell highlighted in your own text: the dashes, the stock opener, the "it's not X, it's Y" line, the announced noticing, the engagement bait at the end, the hashtag wall, the flat rhythm. These are the rules our own writing is held to, made public. No account, no email needed to see the score.
No post to check? Drop screenshots of your LinkedIn instead and see posts in your best voice, held to these same rules.
The nine tells
Dashes
An em dash or en dash used to join clauses. The most recognised tell; commas, periods and colons do the same work.
Stock opener
"Curious...", "Hot take", "Unpopular opinion", "Let that sink in" and their cousins: lines readers have seen a thousand times.
The contrast line
"It's not about X. It's about Y." A rhythm that sounds like insight and carries none.
Announced noticing
"I've been thinking", "I noticed something", "It struck me": narrating the act of having a thought instead of saying it.
Engagement bait
"Agree?", "Thoughts?", "What's your take?", or any closing question by habit. The house ends on the last concrete point.
Hashtag block
Five or more hashtags, or hashtags inside the sentences.
Emoji patterns
Emoji as bullets, or more than one every forty words.
Flat rhythm
Six or more sentences of nearly the same length, or a run of three-word chops. People vary.
Listicle plus one-liner
A numbered list with a one-line kicker, or five single-sentence paragraphs in a row.
Questions people ask
How do I make a LinkedIn post not sound like AI?
Start on the specific thing: the number, the client, the mistake, the day. Cut the stock opener and the closing question. Use commas, periods and colons instead of dashes to join clauses. Let sentences vary in length. Say the Y part of an “it's not X, it's Y” line on its own, with the specific behind it. Then paste it here and see what is left.
Why does AI writing use em dashes so much?
Models learned from a great deal of edited prose in which the dash is a graceful way to join two clauses, and they reach for it far more often than people typing on a phone or at speed do. Readers have now seen enough of it that one dash makes them look for the rest of the tells. The fix is a comma, a period or a colon.
What are the tells that a post was written by AI?
Dashes joining clauses, a stock opener (“Curious...”, “Hot take”, “Unpopular opinion”), the “it's not X, it's Y” contrast, announced noticing (“I've been thinking”), engagement bait or a closing question by habit, a wall of hashtags, emoji as bullets, a flat sentence rhythm, and the listicle with a one-line kicker. This page checks for all nine. None of them proves who wrote the text; they are what readers have learned to skip.
Does this tell me whether a post was written by AI?
No. It reads for the tells readers have learned to skip. A person can write every one of them and a model can avoid all of them. The score says how many tells a post carries, not who wrote it.
Built by PostAgent. The agent that writes the founder's LinkedIn is held to these rules on every post, so this check is also its own QA. The posts are public.