Will people be able to tell a post was written with AI?
They can tell when it's generic. Posts built from your own stories and numbers, edited in your words, read like you because the substance is yours.
Guide
How to get AI to write LinkedIn posts in your own voice
Short answer: show the AI real examples of your writing, describe your voice in plain rules, give it raw material (a story, an opinion, a number) rather than a topic, and edit every draft. AI sounds generic when it's asked to invent things; it sounds like you when it's shaping what you already think.
By the Meg team · Last reviewed 24 October 2026
1. Collect your own writing
Gather 10 to 20 things you've written: past LinkedIn posts, X posts, newsletter intros, even long Slack messages. Include the posts that got the most comments. If you haven't posted much, record yourself explaining your work for five minutes and use the transcript. Spoken explanations are often closer to your real voice than polished writing.
2. Describe your voice as rules
“Friendly and professional” tells an AI nothing. Concrete rules do. Fill in a profile like this and keep it somewhere you can paste it:
Voice profile - Sentences: short. One idea per line. No semicolons. - Words I use: "honestly", "the boring part", "here's what worked" - Words I never use: "game-changer", "unlock", "delve", "in today's fast-paced world" - Formatting: line breaks between thoughts, no hashtags, at most one emoji - Stance: practical, a bit contrarian, always admits what went wrong - Audience: early-stage founders who build B2B software - Topics: pricing, first customers, shipping fast
3. Give it raw material, not a topic
“Write a post about pricing” produces the same post everyone else gets. Give it something only you have:
Using my voice profile and the example posts above, draft a LinkedIn post. Raw material: we raised our price from $19 to $49 last month. Signups dropped 10%, revenue went up, and two customers told us the old price made them doubt we'd last. Angle: underpricing signals that you might not be around. Keep it under 150 words. End with a question, not a call to action.
4. Edit out the tells
Cut the throat-clearing
Delete the first line if it's a generic hook. Start with the specific thing.
Replace vague with specific
Swap “a lot of customers” for the real number, and “recently” for the month.
Remove words you'd never say
If it isn't in your voice profile, it doesn't belong. Add repeat offenders to the never-use list.
Read it out loud
If you wouldn't say it to a peer over coffee, rewrite it until you would.
5. Learn from what lands
Once a month, look at which posts got comments (not just likes) and add one or two of them to your example set. Over time the examples, and the drafts, drift toward what your audience responds to.
Doing this in a general AI chat app
Most AI chat apps let you save standing instructions or create a project with files. Put your voice profile and example posts there so you don't paste them every time, then use step 3 for each new draft. The catch is upkeep: you have to keep the examples current and remember what you've already posted.
Where Meg fits
Meg's AI ghostwriter does steps 1, 2, and 5 for you. It reads your website and public LinkedIn, X, and Instagram posts, builds a style profile you can edit, and learns from engagement on what you publish. For step 3, Spark turns one raw thought into nine drafts, three for each of LinkedIn, X, and Instagram. Editing and posting stay with you.
Questions
How many example posts does the AI need?
Ten good examples beat fifty average ones. Include your best-performing posts and a few that sound most like you.
Should I let AI post automatically?
We don't recommend it. Reviewing every post is how you catch generic phrasing and keep your name on things you actually believe.
Related
Skip the setup
Meg builds your voice profile from what you've already written. Start free.