How to Write LinkedIn Posts in Your Own Voice With AI

AI posts sound generic because the model only knows what the prompt tells it, and most prompts say little more than "professional but friendly". Without your past posts, opinions, stories and audience, the model falls back to the most average phrasing for the topic, which is exactly what readers have learned to scroll past.

A language model predicts likely text. Ask it for "a LinkedIn post about hiring" with no other context and it will produce the most typical hiring post it can, complete with the familiar opening line and the familiar lesson. That is not a model failure; it is the model doing what it was asked.

AI use is now normal among marketers, so the average AI post is everywhere. HubSpot's 2026 research reports that most marketers use AI for content creation, and Content Marketing Institute found that very few B2B marketers have high trust in generative AI output. The gap between "everyone uses it" and "few trust it" is the editing work this guide is about.

A voice profile is a short, structured record of how you write and what you believe: sample posts you are proud of, recurring phrases, sentence length, topics you own, opinions you will defend, stories you reuse, words you never use, and claims you will not make. It gives an AI model something specific to imitate and limits to respect.

Style is the smallest part of voice. Two founders can both write short sentences and still sound nothing alike, because what makes a voice recognizable is the point of view behind it: what you notice, what you disagree with, and which examples you reach for. A useful brand voice model captures both.

1. 01

   Collect 10 to 20 of your own posts

   Pick ones that performed well and that you would still sign today. These are the strongest signal a model can get.

2. 02

   Write down your opinions

   List positions you hold about your market, including the unpopular ones, in one or two sentences each.

3. 03

   Log your stories and examples

   Customer moments, mistakes and decisions you can tell again with real detail.

4. 04

   Set boundaries

   Words you never use, topics you avoid, and claims you will not make without a source.

5. 05

   Describe the reader

   Who you write for, what they already know, and what they are trying to get done.

6. 06

   Review and update monthly

   Add new posts and opinions so the profile keeps up with how you actually write.

Feed the model raw material, not a topic. Paste the call note, the customer objection, the product decision or the half-written idea, tell it which goal the post serves, and ask for a draft that uses only those facts. Grounded drafts need less editing and are much less likely to invent numbers or quotes.

The fastest way to a generic post is a one-line topic. The fastest way to a usable draft is a specific input: "Here is what a customer told us on Tuesday, here is what we changed, write a post for operations leads about why." The founder-led content system guide covers where that material comes from.

Ask the model to mark anything it added that was not in your notes. That one instruction makes fact-checking quick and catches invented statistics before they reach your audience.

Decide on a disclosure policy and apply it consistently. Pew found most Americans say it matters to them whether content was made by AI or a person, and many are not confident they can tell. Many companies with AI guidelines cover disclosure. At minimum, never present AI-invented experiences or quotes as your own.

There is no single rule for disclosure on personal LinkedIn posts, but readers do care. Pew Research Center found that a large majority of Americans think it is important to be able to tell whether text was written by AI or a person, while about half are not confident they could detect it. CMI found that nearly half of companies' AI guidelines cover disclosure.

The practical line most founders use: AI can help organize and phrase your own ideas, the same way an editor or ghostwriter would, but the experiences, opinions and numbers must be real and yours. LinkedIn takes a similar line with its own Post proofreader, which it describes as an assistant that edits what you write rather than writing for you.

Read the draft aloud, cut the first line if it sounds like everyone else's, replace general claims with one specific example, remove words you would never say, check every fact against your notes, and end on a real question or position. If a sentence could appear under anyone's name, rewrite it or delete it.

Editing is where the voice is won or lost. A quick pass for rhythm, then a slower pass for substance, catches most problems. Watch for the tells readers recognize: tidy three-part lists, inflated adjectives, and conclusions that restate the opening.

In ContentOS, voice-aware drafting drafts from your own imported posts and the opinions and stories stored in Creator Brain, and every draft goes through your review before it is scheduled. The use cases page shows how founders and small teams set it up.