Ghostwriter vs AI vs writing it yourself: who should write a CEO's LinkedIn?

What are the three ways to write your LinkedIn posts?

Every CEO who wants to show up on LinkedIn ends up with the same question: who writes? There are three answers:

1. Write it yourself. You sit down, think and write. Maybe with help from someone in marketing.
2. An AI tool. ChatGPT, Claude or a dedicated LinkedIn post generator. You give it an idea, it drafts, you edit.
3. A ghostwriter. A person who learns you, your company and your audience, and writes in your name and your voice. You approve every post before it goes live.

I am a ghostwriter, so I have an obvious interest. Still, the honest answer is that no single option is right for everyone. There are CEOs I tell on the first call: "You don't need me." This article is about exactly that: when each option is right, and when it isn't.

Ghostwriter vs AI vs writing it yourself: how do they compare?

The first two rows matter most. Almost the whole decision comes down to how much time you have, and whether you already have a writing voice of your own.

When is writing it yourself the right choice?

When you have fixed time and someone to handle the day-to-day. If a CEO tells me, "I have an hour at the start of the week when I sit and write, and an hour at the end of the week when I write about what happened, what I did, what I concluded and what interests me," and they have a marketing team that can write and post about events during the week, I tell them: you don't need me.

Don't underestimate that time. In Orbit Media's 2026 survey of 1,042 content marketers, the average blog post took 3 hours and 20 minutes to write. A LinkedIn post is shorter, but the thinking, the wording and the editing are the same work. If you write yourself, you also need a system, or you get stuck by week three. I described mine in the content waterfall: one idea, 28 pieces of content.

When is an AI writing tool enough?

When you already have a voice. If you have your own way of speaking and writing, and you have written a lot in the past, you can let an AI model learn how you write, and it will produce something good enough. Time matters here too: if you have one to two hours a week to sit with the model, steer it, fix it and write, it fits you. If you don't, better not to. The results will be poor, and people will spot the AI in a second.

Company size matters as well. At small companies, an AI tool is often the starting point, because they are not ready to hand writing to an outside writer. At large companies it is the opposite: they usually won't accept publishing AI output under the CEO's name, and they hand it to a writer.

What do AI tools get wrong?

They sound like everyone, because they really do write like everyone. You can spot an AI post by its sentences, its phrasing, the long em dash and all the familiar tells. And it is no longer a small problem:

• Most long posts already look like AI. Originality.ai checked 5,000 long-form posts (100+ words) from July 2026: 81.2% were classified as likely AI, up from about half in late 2024. Pangram scanned over a million posts in 2026 and found that more than 40% of long-form LinkedIn posts were fully AI-generated. The detectors disagree on the exact number, but they agree it is a lot.
• People notice, and tune out. In a Bynder survey of 2,000 consumers in the US and UK (2024), 50% correctly identified AI-written copy, and 52% said they feel less engaged with content they suspect was written by AI.
• Trust drops. In a Meltwater and YouGov study of about 10,000 consumers in seven markets (April 2026), 32% said they would trust a brand less if its content was AI-generated, and only 15% said they would trust it more.

And when everyone publishes the same text, what stands out is what only you have: the opinion, the story, the customer who called yesterday. No model knows that without you.

Is it okay to use AI to write LinkedIn posts?

As an editor, yes. As a writer that replaces you, it is risky. LinkedIn itself draws the same line:

• On July 30, 2026, LinkedIn added a "Seems like AI slop" option for posts, to better identify low-quality AI content and show less of it in recommendations from outside your network (TechCrunch).
• According to LinkedIn's chief product officer, Hari Srinivasan, more than a million people used the option in its first two weeks. LinkedIn made clear that using AI to refine the language of what you wrote is not the target (ContentGrip, August 2026).
• In the same update, LinkedIn pulled its own "enhance your post" AI writing feature and replaced it with a proofreading tool that doesn't change your voice (TechCrunch).

That is the clearest signal: the platform is not against AI. It is against posts with nobody behind them. If the idea, the opinion and the story are yours and AI helped with the wording, that is fine. If AI invented the post and you only clicked "Post", readers will notice.

When is a LinkedIn ghostwriter worth it?

When you don't have the time. A CEO who has no free moment during the week, who can't sit down and read on X or LinkedIn what competitors are doing and what is happening in the market, let alone write about it, is exactly who needs me.

And silence has a cost. Showing up on LinkedIn takes consistency: according to Richard van der Blom, who analyzed 1.3 million posts from 50,000 creators for his 2026 Algorithm Insights report, the optimal frequency is two to four posts a week. And a personal profile reaches more people than a company page: 1,006 median impressions per post against 659 (AuthoredUp, 2026). I wrote about that in the guide to founder-led marketing.

Some of my clients came to me after AI tools, and some after another ghostwriter. Those who came after AI said their posts had no traction and no momentum, and the right people never showed up. Those who came after another ghostwriter felt it wasn't them: generic, like everyone else. Once we started working together, they felt the content was authentic, personal and sounded like them. That is how I work as a LinkedIn ghostwriter for CEOs: every post is written by hand, and nothing goes live without your approval.

How does a ghostwriter capture a CEO's voice?

Everyone asks this, and rightly so. A ghostwriter who doesn't learn you will write generic posts, and then you might as well use AI. This is what the first weeks look like with me (sometimes it takes less, depending on how much time the CEO gives):

1. I join meetings. I learn the company, hear how the CEO talks and which words they use to describe what the company does.
2. I learn the customers. What interests them, what hurts them, and what makes people want to work with this CEO.
3. I interview. About sales calls, customers, what interests the CEO and what they won't say.
4. I write sample posts. The CEO reacts, corrects me, and together we learn how they want to sound.

After that, the ongoing work is short. 30 to 60 minutes a week on content, a call or WhatsApp voice notes, depending on the CEO (outbound, if it is included, has its own time). I write almost all the posts myself, and the CEO approves them. For personal posts, like a story from their life, we get on a video call, I write a draft, and they refine it.

An example from a normal week: the CEO of a cybersecurity company sent me an article about a new breach and asked for a response, a counter-article with a fitting case study from his company. Within two hours he had an article in his own tone, with the example from his company, ready to publish. That is the difference between someone who knows you and a tool that knows the internet.

What about an AI trained on your own writing?

There is a fourth option, in between: an AI model that learns your tone, your company, your customers' pains and what you solve, and drafts with you in your voice. I built a product like that, Velut AI, out of my work with CEOs.

But it is worth being honest about who it fits: CEOs who have the time and are willing to sit and write with an AI tool built for this work. Most CEOs won't be able to sit and write even with that tool, so they prefer to work with me. And for managed clients, I write by hand, not with the model.

How do you choose? Four questions

1. Do you have one to two hours a week, every week? If not, writing yourself and AI will both break down. Ghostwriter.
2. Do you already have a writing voice, and a lot of text you wrote? If yes, an AI that learned you may be enough. If not, it will invent a generic voice for you.
3. Who checks before it goes live? If it is only you, and you are busy, AI tells will reach the feed. A ghostwriter is a second pair of eyes.
4. What is the goal? For a basic presence, any of the three can work. If the goal is for the right decision-makers to reach out to you, your voice is the whole story.

The bottom line

There is no single answer. If you have time and a voice, write yourself, and use AI as an editor if you like. If you have a voice, plenty of past writing and one to two hours a week, an AI that learned you can work. And if you don't have a free hour a week, don't publish generic AI drafts under your name. That is what a ghostwriter is for. If that is you, this is how I work: I learn you, write by hand in your voice, and you approve every post.