It's late 2023. A SaaS founder, let's call him Sawyer fires his content team. Two writers, one strategist. Combined salary: about £120K a year.

His reasoning? ChatGPT can do it. The blog posts look the same. The LinkedIn carousel templates look the same. The email sequences look close enough. And the AI doesn't take bank holidays off or ask for a mental health day because someone was rude to it on Slack.

Sawyer isn't stupid. He's pragmatic. He's running the maths and the maths says: why pay humans £120K to do what a £20/month subscription does in 40 seconds?

So he does what a few hundred thousand other founders did between 2023 and 2024. He replaces the writers. Fully automated. Blog posts generated Monday, scheduled Tuesday, published Wednesday. LinkedIn content is batch-produced on the first of every month. Newsletter drafted, polished, sent.. all without a human fingerprint anywhere near it.

For about four months, the dashboards looked fine. Content volume went up. Publishing frequency tripled. The graphs pointed in the right direction.

Then the graphs stopped moving.

Then they started going backwards.

Here's the thing about AI-generated content that nobody warned these founders about.

It's not bad. That's the problem.

If it were bad.. obviously bad, comically bad, the kind of bad that announces itself you'd catch it immediately and fix it. But AI content isn't bad. It's beige. It's competent, grammatically correct, structurally sound, and completely invisible.

It reads like a press release that shagged a Wikipedia article.

Every blog post has the same cadence. Every LinkedIn hook follows the same pattern. Every email sequence uses the same transitions. It's the content equivalent of a hotel room, technically it has everything you need, but you'd never call it home. You wouldn't remember a single detail about it three hours after checking out.

And here's where the maths gets spicy.

When one founder automates their content, they get an edge. When ten thousand founders automate their content using the same tools, prompts, and models, they all get the same edge. Which means nobody has an edge. The floor just rose, the ceiling didn't move, and everyone's content is now competing in a pool of indistinguishable, algorithmically generated beige.

Sawyer's blog posts were technically better than before. More frequent. More "optimised." But his organic traffic dropped 34% in six months. His LinkedIn engagement went from an average of 47 comments per post (when his team was writing) to 11. His newsletter open rate, which had been hovering at 42%, slid to 28%.

Not because the content was worse on paper. Because it was worse in practice. Because nobody could feel a human behind it.

I've been ghostwriting for founders for years. And I'll tell you something that won't surprise anyone who's been in this game longer than five minutes.

Every time a new tool shows up that promises to replace writers, the same cycle plays out.

𝗣𝗵𝗮𝘀𝗲 𝟭: excitement. "This changes everything."

𝗣𝗵𝗮𝘀𝗲 𝟮: adoption. Companies replace writers, cut budgets, and automate everything.

𝗣𝗵𝗮𝘀𝗲 𝟯: saturation. Everyone's using the same tool. Content starts blending together.

𝗣𝗵𝗮𝘀𝗲 𝟰: correction. Companies realise the human was the differentiator all along. They hire writers back. Usually at higher rates.

We've seen this with content mills. With SEO article spinning. With offshore content farms. And now with AI.

A freelance writing industry analysis from earlier this year put it bluntly: companies that adopted AI content in 2023-24 are now desperate for experts who can craft authentic, meaningful language. The exact skills in highest demand? Executive ghostwriting. Brand voice development. Messaging guides. Content strategy.

Read that list again.

Executive ghostwriting. Brand voice development.

That's not a list of "nice-to-haves." That's a list of things AI fundamentally cannot do, because they all require something AI doesn't have.

A point of view.

PR Daily recently asked comms leaders across the industry what they predicted for 2026. The responses are worth paying attention to.

Duolingo's Director of PR said she expects audiences to start preferring imperfection in content because imperfection signals humanity. The CMO of Scale Venture Partners went further. Her argument, paraphrased: when infinite AI content becomes the standard, the only thing that differentiates you is being unmistakably human. She called it "the human moat."

I wanted to frame that and hang it on my office wall.

Because that's the thing I've been saying to founders for the last three years, except I've been saying it with more swearing and fewer VC-friendly metaphors.

Your voice is not a nice-to-have. It's infrastructure.

The opinions you've been sanding down in Google Docs because they feel "too strong." The stories you've been leaving out because they're "too personal." The way you actually talk when you're three drinks in and someone asks you a question you care about. 𝗧𝗛𝗔𝗧 is the content that cuts through.

Not the template. Not the framework. Not the AI-polished, committee-approved, legally-reviewed, personality-free content that reads like it was written by a toaster with a LinkedIn subscription.

The voice. Your voice. The weird, specific, slightly uncomfortable one that sounds like you and nobody else.

That's the moat.

Sawyer hired a ghostwriter. Not a content agency. Not a freelance platform. A single human being who spent four hours on a call extracting his actual opinions, the ones he'd been too nervous to publish because they were "controversial."

(They weren't controversial. They were specific. Which on LinkedIn is basically the same thing.)

Within three months, his blog traffic recovered. Within four, it exceeded the pre-AI numbers. His LinkedIn engagement climbed back above 40 comments per post and the comments were different. Longer. More substantive. From actual buyers, not bots and fellow content marketers.

The difference wasn't volume. He was posting LESS than the AI had been producing. Three pieces a week instead of seven.

The difference was that every piece sounded like it came from a specific human with a specific perspective and a specific reason for writing it. People could feel that. The algorithm could feel it too .

LinkedIn's 2026 updates explicitly reward what they call "thinking content" over "teaching content." Translation: your opinion now outranks your framework.

Sawyer's AI content was teaching content. Competent, comprehensive, forgettable.

His ghostwritten content was thinking content. Opinionated, specific, his.

Same founder. Same expertise. Same topics. Completely different results.

If you're a founder posting AI-generated content right now, I'm not going to tell you to stop. That would be simplistic and also a bit rich, given that I use AI in my own workflow.

But here's what I will tell you.

AI is brilliant at speed. Structure. First drafts. Research synthesis. Volume.

AI is shite at voice. Perspective. The thing that makes someone stop scrolling and think, "this person gets it." The thing that makes a buyer feel like they already know you before they've ever spoken to you.

And that second list: voice, perspective, recognition, trust, is the list that actually produces pipeline. Not impressions. Not reach. Pipeline. Real humans with real budgets who DM you because something you wrote made them feel understood.

The companies that spent 2024 replacing writers with AI are spending 2026 hiring them back. Not because AI failed. Because AI succeeded at the wrong thing. It succeeded at producing more content. It failed at producing content that mattered.

The pendulum always swings. It swung toward automation. Now it's swinging back toward voice.

The founders who figure that out first will spend the next twelve months building trust while their competitors are still feeding prompts into a machine and wondering why nobody's responding.

→ Your voice is the thing AI can't replicate.

→ Your perspective is the thing that stops the scroll.

→ Your specificity is the thing that converts attention into pipeline.

That's not marketing advice. That's maths.

Here's a question I want you to actually answer, not in your head, on paper.

Go read your last five LinkedIn posts. Read them out loud. Do they sound like you? Do they sound like the version of you that exists at dinner with a friend, or the version of you that exists in a "professional" Google Doc?

If there's a gap between those two versions, that gap is costing you money. Not in some abstract "brand equity" way. In actual pipeline. In actual conversations that aren't happening because your content sounds like everyone else's.

The writers are coming back. The question is whether you'll be the founder who hires one now, or the founder who hires one in six months after watching your competition do it first.

Message me if you want to find out what your voice actually sounds like when it's not cosplaying as a corporate robot.

Or stay beige. Your call.

Written by Sarra, the Ghost, professional ghostwriter, strategist, and marketer hiding in plain sight behind the voices you read every day.

Find me here → linkedin.com/in/meetsarra