How to write LinkedIn posts that don't sound like AI?

AI content has a smell, and your buyers pick it up from across the room. AI content reads polished and empty at the same time, like a brochure wearing human clothes.

The problem is bigger than taste. When a buyer checks your profile before replying to you, your posts are standing in for you. If they read like everyone else's, you look like everyone else, and the reason to talk to you disappears. How a post sounds matters as much as what you post to attract clients, because a good idea in a generated voice still reads as generated.

Why do so many LinkedIn posts read like AI?

Because a lot of it is written by the same AI tools, prompted the same way. Senior people who feel nervous about posting reach for an AI tool to make it safe, and the AI makes it bland. The same rhythm, the same openers and the same three tips land in every feed, so every post blurs into the next.

The Noise Machine runs on this. Generated posts by the hundred, AI comment tools leaving "Great insights!" under everything, engagement pods trading likes to make it all look popular. Everybody is posting, almost nobody is trusted, which is part of the reason so many posts get almost no impressions.

I know the nerves. I couldn't dance, and I was terrified of talking to girls. Then I answered an advert for ballroom and Latin dance instructors that said "no experience required". I learned both the man's and the woman's steps and became an instructor. Nobody could learn those steps for me. Posting is the same. Hand your nerves to AI and you skip the learning, and the cost is a voice that belongs to nobody. If I can't do something, I learn it, then I teach it. Your voice gets learned the same way, one slightly clumsy post at a time.

What gives LinkedIn posts away as AI?

Seven tells give a post away as AI, and you can learn to spot all of them. Some are in the words, some are in the rhythm, and the worst one is what is missing. AI content is a post where a tool chose the words and the shape, and the person only pressed publish.

1. Long dashes everywhere. Fair or not, readers now take them as a sign the writer handed the job to a machine.
2. Words nobody uses out loud. If you would never say "let's dive in" to a client across a table, it does not belong in your post.
3. Big openers and empty closers. A first line about how fast the world is changing, and a last line that asks "Thoughts?" with nothing to think about.
4. Emoji bullets and a row of hashtags. Decoration standing in for something to say.
5. Everything in threes, at the same length. Three adjectives, three bullet points, three short lines in a row, every sentence the same size.
6. The constant flip. Every other sentence turns on "it isn't this, it's that", or asks a dramatic question and answers it in the very next line.
7. Nothing only you could know. No names, places, dates or numbers. No opinion anyone could disagree with. Nothing that happened to you this week.

That last one matters most. You can clean up every word and still sound generated if there is nothing in the post that only you could have written.

Can people tell when your LinkedIn posts are written by AI?

Yes, more often than you'd think. Your buyers read LinkedIn every day, and they've seen enough AI posts to recognise the shape before they reach the end of the first line. They may not be able to say why. They just feel that nobody was home when it was written, and they scroll on.

The giveaway is rarely one word. It's the whole pattern: the safe opener, the tidy list, the question at the end that nobody needs to answer. One of those on its own is fine. All of them together, in a post with no moment in it, reads as AI to almost anyone who spends time on LinkedIn.

Does LinkedIn penalise posts written with AI?

As far as I know, LinkedIn hasn't said it punishes a post simply because AI helped write it. The penalty comes from your readers. AI posts tend to get skimmed, not answered, and a post nobody stops for or talks about doesn't travel far. The tool isn't the problem. The empty post is.

So don't worry about a detector. Worry about the buyer who checks your profile, reads three posts in a row in the same AI voice, and decides there's nobody behind them worth talking to. That's the penalty that costs you work.

So how do you write LinkedIn posts that sound like you, not AI?

Write posts that sound like you by starting with three things only you have: a moment from this week, what it cost, and an opinion someone could push back on. Then say it out loud before you type, and let AI tidy the shape, never the story.

Here is the prep, on the back of an envelope. A moment: something that happened to you or a client this week. A cost: what it took, in time, money or pride. And an opinion: what you think about it that someone might push back on. If you have those three, the post almost writes itself. If you don't, no amount of editing will save it.

Then speak it before you type it. Say the post out loud into your phone the way you would tell it at a “braai” (barbecue), tongs in one hand, to a client who asked what you have been up to. Spoken words come out in your rhythm, with your quirks, and that is the hardest thing for AI to fake. My own process for anything written is to speak it, let the machine help frame it, then panel beat it by hand before it goes anywhere. Started by a human, polished with AI, finished by a human.

How do you edit an AI draft so it stops sounding generated?

Edit in six passes, each with one job: find the buried hook, swap adjectives for facts, cut the padding, remove the tells, rewrite the first three lines by hand, and read it aloud. Hunting for one thing at a time is slower than one big read, and it catches far more.

1. Find the buried hook. The real opening is usually a few lines down, where the writer stopped warming up and just said it. Cut everything above it.
2. Replace adjectives with facts. Not "a big result" but the number. Not "fast" but "in under two hours". A vague word usually means the draft was never given a fact.
3. Cut the padding. Delete every line that announces what is coming next. If a thing is clear, you don't need to say so.
4. Remove the tells. Go through the list above. Swap anything you would not say aloud for what you would actually say.
5. Rewrite the first three lines by hand. AI can write something correct. Only you can write what you felt.
6. Read it aloud. If any line sounds like nobody talks that way, it is not done.

A little roughness is fine. I am not suggesting you add spelling mistakes, but when something reads as too perfect, people sense it. The same goes for DMs, where a templated message gives you away even faster, as I cover in following up after silence without sounding pushy.

What does an AI LinkedIn post look like before and after a rewrite?

An AI post before a rewrite looks tidy and could have been written by anyone. Here is a made-up example, so you can see the edit. Sam is an accountant who does not exist, and this is what a tool gives him when he asks for a post about cash flow.

Before: "In today's fast-paced world, cash flow matters more than ever. Here are three tips every business owner needs to know. Plan ahead. Track everything. Review often. It isn't about working harder, it's about working smarter. Thoughts? #CashFlow #SmallBusiness #Growth"

Now the same post, rewritten from a moment Sam had this week.

After: "On Tuesday a client rang me in a panic. A £9,000 VAT bill was due on Friday and nobody had put the money aside. We found it, after two phone calls and a very awkward chat with his bank. Now 20% of every invoice he sends moves into a separate account the day it is paid. Boring. It has saved him twice this year. Do you move your tax money on the day, or at the end of the quarter?"

Look at what changed. The after version has a day, a sum, a cost and an opinion. It has one idea instead of three tips. It ends on a question a reader can answer in a few words. And none of it could have come from AI, because AI was not on the phone on Tuesday.

Should you stop using AI for LinkedIn altogether?

No. AI is a word processor with a fancy badge. It is good at structure, tidying and spotting what is missing. It is bad at being you. Use it for the shape, never for the story, the opinion or the opening.

And keep it away from your engagement. AI comments and pods make your profile look busy to strangers and fake to the people who matter. Whether engagement pods help or hurt your LinkedIn reach goes into why. If time is the real problem, how busy founders post consistently shows the monthly routine.

How do you use AI on LinkedIn posts without losing your voice?

Use AI on LinkedIn posts as an editor, never as the author. Give it your own words first, a voice note or a rough draft, and ask it to tidy the structure without changing your phrases. Then take its suggestions back by hand. You stay the writer, and AI does the dull bits.

Here are four prompts that help without taking over. Paste your transcript or rough draft first, then ask AI one of these:

1. "Suggest three ways to order this. Don't change my wording."
2. "Which sentences sound generic or could be cut?"
3. "What would a sceptical buyer ask after reading this?"
4. "Mark every word I'd never say out loud to a client."

Notice what's missing from that list: "write me a post about". The moment AI chooses the idea, the post stops being yours.

What should you never let AI write on LinkedIn?

Never let AI write the parts of LinkedIn that carry your judgement: the first three lines of a post, any story, any opinion, your comments on other people's posts and your direct messages. Those are where a buyer decides whether you're worth a conversation, so they have to be yours, word for word.

Comments matter most of all. An AI comment under a buyer's post is easy to spot, and it tells them you didn't really read it. Two honest sentences you typed yourself beat a paragraph of generated praise every time.

What does it look like when our team writes your LinkedIn posts?

You talk, we type. Our content team interviews you once a month, and a person writes every post from what you said, in your words. AI sits behind the scenes for tidying, never in front. Never templated ghostwriting. You approve every post before it goes out, and if a line does not sound like you, it changes.

Content Starter is $550 a month for four posts from one short interview. Content is $1,600 a month for 12 posts and a newsletter from one 30-minute interview, plus daily engagement. Month to month, no contract.

Who is this not for?

If you want a high volume of posts with no input from you, we are the wrong people, because there is no way to sound like you without you. If you expect posts to bring in clients on their own, they won't, for anyone. And if your referrals are still pouring in, you don't need us.

Start with a moment only you saw and an opinion you will defend. Speak it, shape it, then make it yours by hand. Do that and nobody will ever say your posts sound like AI.