One thing I’ve started telling my content marketing clients more often lately: don’t send me AI-written first drafts.

The issue isn’t AI itself, as I use it myself for some writing tasks. The issue is what gets lost when you start with AI. I’ll explain.

Part of my work as a content marketing strategist involves helping executives turn ideas into articles, newsletters, LinkedIn posts, and other forms of thought leadership.

The process is (or rather, was) usually simple: we discuss a topic, and they send me notes, voice messages, bullet points, half-finished thoughts, or rough drafts. My job is to help shape those ideas into something people will want to read.

Increasingly, however, I’m receiving three polished pages generated by AI instead of shorter, messier drafts. And you might think that’s better – more coherent, structured, flawless, easier to edit. It’s better, but it’s not. Because it’s missing something essential I absolutely need for content to work: your tone of voice.

AI genuinely helps people write

According to my 2026 Writing at Work survey, AI is now part of the writing process for most professionals.

The benefits people reported include:

• 44.1% say they write faster
• 41.2% say their writing is more structured
• 20.6% say their writing is clearer
• 19.1% say their writing sounds more confident or credible

Overall, 40.6% described AI’s impact on their writing as mostly positive.

This isn’t surprising, as many writing challenges are execution ones: getting started, organizing ideas, finding the right format, and polishing a draft. AI is often very good at these tasks.

The problem is that it can also remove individuality. As a result, a lot of content starts sounding the same – you might notice this on LinkedIn, for example.

AI efficiency comes with trade-offs

The same survey revealed that:

• 27.9% say they rely on AI more than they would like
• 26.5% say their writing sounds more generic
• 25.0% say they spend too much time editing AI output

In addition, one of the most commonly mentioned dislikes in workplace writing was content that feels “over-polished” or artificial.

This is where the main issue is. AI is excellent at producing polished writing. Which means it reproduces patterns, phrases, and formulas that AI thinks make for good writing. Unfortunately, that writing doesn’t always sound, well, human.

For routine communication, this may not matter much. For thought leadership, personal branding, and expertise-based businesses, it matters a lot.

Messy drafts are valuable

One reason I prefer receiving rough notes instead of polished AI drafts is that messy writing contains useful information. I want to see how someone naturally explains an idea:

• which words they repeat
• which analogies they use
• how direct or cautious they are
• where they get more passionate
• what they choose to emphasize
• how they describe their business/services

AI may choose to edit out many of these particularities to make the text fit into a proven formula. Ironically, the parts that get “cleaned up” are often the most valuable clues about their voice.

So, I started telling my clients: The rough draft doesn’t need to be good. Don’t worry about sending me a messy text. In fact, I prefer it that way.

Using AI without losing your voice

1. Write the first layer yourself

Start with your own thoughts. This doesn’t mean writing a perfect first draft.

Write:

• bullet points
• a rough outline
• a few paragraphs
• record a voice note and transcribe it.

Create raw material that comes from you rather than from a prompt. AI works much better when it has something authentic to build on. So do content writers.

2. Use AI as an editor

Most people immediately ask AI to generate content on theme X or Y. I often get better results by asking it to critique content instead. Ask questions like:

• How would you tighten this text without losing the original tone of voice?
• Thinking like an editor, what suggestions for improvement do you have?
• Where are arguments weak and what needs more evidence?

These prompts improve the quality of thinking without replacing it.

3. Add specifics that AI cannot invent

AI-generated writing is often generic and lacks credibility. Specificity is one of the fastest ways to make writing more credible. Add:

• personal examples from your work
• lessons learned and mistakes made
• observations from conversations
• numbers and results (especially your own)

4. Read the draft (aloud)

Finally, after going through AI-editing, re-read the text – if you can, read it out loud.

As you read, look for phrases you would never naturally use in conversation or in writing. Cut or rewrite them.

If something feels unfamiliar, strange, or artificial to you (and your voice), the reader will pick up on it.

5. Create a personal style guide

Record yourself or archive older texts. Ask AI to analyze your voice and find particularities in your voice:

• expressions you tend to use
• certain sentence structures
• metaphors or analogies you naturally bring up

These are often part of what makes your writing recognizable. Let’s call it your style (which AI tends to flatten). Use the analysis to write a style guide for your voice.

6. Don’t optimize everything

I’m all for going through multiple rounds of editing before publishing anything. This has been my process for years (10, 20, who’s counting?)

But this is something you can overdo. And, if you use AI for editing, you will definitely overdo it at lightning speed.

AI will polish mercilessly, to the point that it sounds artificial.

Consider only editing parts of your writing. Maybe get it to tighten up a report that’s getting too long or to structure an article where you feel stuck. But keep your emails and social media posts fresh. Use a tool like Grammarly to ensure you avoid typos and grammar errors, but otherwise, go with an AI-free flow every once in a while.

Trust me, your own voice is far more interesting than you think and will stand out in times when everyone is tempted to sound correct but bland.