How I Write With AI (Without Losing My Soul)

Here’s something that might surprise you: 98% of my LinkedIn content these days is AI-generated.

Yet I’ve built an audience of over 25,000 people, get consistent engagement on my posts, and regularly receive inbound leads from my content. Before you roll your eyes or call me a fraud, let me explain why this isn’t what you think.

What’s Broken in AI Writing?

I’ve been watching the AI writing landscape for months now, and I keep seeing the same three broken approaches over and over again.

1. The Lazy Prompters

First, there are the lazy prompters who throw out something like “write me a LinkedIn post about productivity” and then get frustrated when the AI spits out generic, fluffy garbage that reads like it was written by a committee of corporate consultants. They try it once, decide AI writing is useless, and go back to doing everything manually.

2. The Content Millers

Then, on the other extreme, you’ve got those content mill people. These are the ones who think they’ve cracked the code by using AI to pump out hundreds of blog posts, thinking quantity equals success. They’re flooding the internet with low-quality content that helps nobody and makes the web a worse place for everyone.

3. The Pro Writer Sceptics

Finally, there are the professional writers who dismiss AI altogether and actively discourage others from using it in their craft. I get it — I’m a professional writer too, and I understand the fear that AI will somehow diminish the art of writing. But this approach is not just counterproductive, it’s putting these writers at a massive disadvantage.

The real problem isn’t AI itself. It’s that people haven’t been taught how to use AI properly for writing. They either fear it completely or expect it to produce brilliant results with minimal input from them. Neither approach works.

The right way to write with AI is a hybrid approach where you and the AI work together as partners. You bring the ideas, insights, and judgment. AI handles the structure, flow, and grunt work of turning your thoughts into polished content. When done right, this collaboration produces better writing than either human or AI could create alone.

AI as a Writing Partner, Not a Replacement

Before I show you the specific tactics, I want to share the mindset shift that changed everything for me. Most people think about AI writing all wrong because they’re asking the wrong question.

The question isn’t “Can AI write better than humans?” The question is “Can AI help me express my ideas more effectively?” That subtle difference changes everything about how you approach the technology.

Much of good writing is actually good thinking. If you can nail the thinking part, AI becomes incredibly powerful at handling what I call the “grunt work” of writing — sentence structure, paragraph flow, transitions, formatting, even optimisation for search. These are important, but they’re not where your unique value comes from.

Think about it this way. When typewriters were invented, nobody worried that they would make writing less authentic. When word processors came along, writers didn’t panic that spell check would somehow diminish their craft. AI is just the next evolution of writing tools, and like those previous innovations, it’s incredibly good at handling the mechanical aspects of getting words onto a page.

What AI can’t do is have your experiences, form your opinions, or generate your insights. It can’t understand your industry the way you do or connect with your specific audience based on years of relationship building. That’s all you.

So here’s my framework: I handle the thinking, AI handles the execution.

I bring the ideas, the personal anecdotes, the industry insights, the controversial opinions that make content worth reading. AI takes those raw thoughts and helps me structure them into something that flows well, reads smoothly, and hits the right notes for the platform I’m writing for.

And here’s something that might surprise you: I’ve been going back and improving some of my old blog posts that were written 100% manually, using Claude and ChatGPT. The AI-enhanced versions consistently have better structure, smoother flow, and clearer narrative arcs than my original manual writing. The ideas are exactly the same, but AI helped me express them more clearly.

That’s when it clicked for me. AI isn’t making my writing less authentic — it’s helping me be more authentic by removing the barriers between my thoughts and clear expression of those thoughts.

My LinkedIn Process: From Voice Note to Viral Post

Let me walk you through exactly how I create LinkedIn content that actually performs.

It all starts with a messy voice note

My LinkedIn posts don’t begin with me sitting down at a computer trying to craft the perfect opening line. They start when I’m walking my dog, drinking a morning coffee at my local cafe, or lying in bed thinking about something that happened during my day.

When I have an insight or observation I want to share, I immediately voice dictate it into a dedicated automation I built. And I mean immediately — whatever messy, unstructured thoughts are in my head.

Here’s what a typical voice note might sound like: “I was talking to this client today and they told me they’re using seventeen different AI tools but they can’t point to a single business outcome and it made me think about how most companies are treating AI like Pokemon cards, gotta catch them all, but they’re not actually getting better at anything specific…”

Notice how rambling and unstructured that is? That’s intentional. At this stage, I’m not worried about hooks or formatting or calls to action. I’m just trying to capture the core insight while it’s fresh.

The AI translation layer

Behind the scenes, I have an extremely comprehensive prompt that takes these messy voice notes and transforms them into properly structured LinkedIn posts. This prompt is built on two key components: examples of my own writing style and proven frameworks for high-performing LinkedIn content.

The prompt knows how to extract the core insight from my rambling thoughts and structure it with a compelling hook, properly spaced body text that tells a story or makes an argument, and an engagement question that actually gets people to comment.

But here’s the key part: the AI isn’t generating the ideas. It’s taking my ideas and giving them structure. The Pokemon card observation about AI tools? That’s mine. The observation about clients not being able to point to business outcomes? Also mine. AI just helps me present these thoughts in a way that performs well on LinkedIn.

The human quality gate

A lot of people think automation means “set it and forget it.” I never auto-publish anything. Ever.

Once the AI generates a draft, I read through it carefully. I’m checking for accuracy, making sure the tone feels right, and ensuring it actually captures what I was trying to say. Sometimes I’ll adjust the hook, tweak the ending, or add a detail that makes the story more specific.

This review process usually takes just a few minutes because my AI drafts are typically 90% there. But that final 10% of human judgment is what keeps the content authentic and ensures I’m comfortable putting my name on it.

The results speak for themselves: over 25,000 followers, consistent engagement, and regular inbound leads from people who connect with the insights I’m sharing. The ideas remain 100% mine — AI just handles the heavy lifting of turning scattered thoughts into coherent, well-structured posts.

My Blog Writing Process: Deep Research + Personal Insights

Blog posts require a different approach than LinkedIn content because they’re longer, more detailed, and need to compete in search results. But the core principle remains the same: I handle the thinking, AI handles the execution.

Research phase

Every blog post starts with research. Before I write anything, I want to understand what’s already ranking for the topic I’m covering.

I’ll typically use Perplexity or ChatGPT o3 to pull comprehensive research on my topic. But I don’t just look at the top Google results — I also dig into Reddit discussions, forum conversations, and industry-specific communities to understand what questions people are actually asking and what gaps exist in the current content.

The goal isn’t to copy what’s already out there. It’s to understand the landscape so I can identify where my perspective and experience can add unique value.

Outline development: where the magic happens

Once I understand what’s already been covered, I ask my AI tool to help me build an initial outline based on the gaps I’ve identified and the frameworks I want to share.

Then I enrich that outline by voice dictating my thoughts about each section. This is where I pour in my personal insights, client stories, opinions, and experiences that make the content uniquely mine.

This is the most important step in my entire process because it’s where authenticity gets baked in before AI starts building out the sentences and paragraphs.

AI enhancement: structure and flow

I take that enriched outline and feed it into another comprehensive prompt that includes examples of my writing style and proven blog post structures. What emerges is a full blog post that captures my ideas and experiences but presents them with better structure and flow than I might achieve writing manually.

Quality control: the final filter

Just like with LinkedIn, I never publish anything without human review. With blog posts, this is even more critical because these pieces represent my expertise and position me as an authority in my field.

I read through every post to check for accuracy, ensure the examples are relevant, and verify that the tone aligns with my brand. I’ll often add specific details to stories, adjust technical explanations, or refine the call-to-action based on my current business goals.

4 Keys to Authentic AI Writing

After months of refining this process, I’ve identified four non-negotiable principles that separate authentic, valuable AI-assisted content from the generic garbage flooding the internet.

1. Start with your own ideas

AI should amplify your thinking, not replace it. The best AI-generated content starts with genuine insights, personal experiences, or unique perspectives that only you can provide. The Pokemon card observation didn’t come from a prompt about business strategy — it came from a real conversation with a frustrated client. That specificity is what makes content worth reading.

2. Build custom prompts, don’t use generic ones

Most people go wrong here because they use prompts they found on Twitter or bought in some course. I spent weeks studying my own writing style, identifying patterns in how I structure arguments, the type of examples I use, and the tone I naturally adopt. Your prompt is essentially teaching AI to write like you, not like some generic “professional writer.”

3. Maintain human oversight, always

Never auto-publish anything. Your name goes on that content, so your standards need to apply. The review process also keeps you engaged with what you’re publishing and ensures you can stand behind every word.

4. Focus on value, not volume

The biggest trap I see people fall into is using AI to dramatically increase their content output without increasing the value they’re providing. Quality always beats quantity. I’d rather publish one genuinely helpful AI-assisted post per week than seven mediocre ones.

The Authenticity Paradox

I discovered something counterintuitive — using AI assistance to better express my authentic ideas often results in more authentic content than struggling to write everything manually.

When I’m wrestling with sentence structure or trying to remember the best way to transition between ideas, I’m not fully focused on the insights I want to share. AI handles those mechanical aspects, which frees me to concentrate on what really matters: the thinking, the experiences, and the perspectives that only I can provide.

The goal isn’t to hide that you’re using AI. The goal is to use AI so effectively that your authentic voice comes through more clearly than it would otherwise.

Building Your Own AI Writing System

The writers who thrive in the AI era won’t be those who reject the technology or those who rely on it completely. They’ll be the ones who learn to dance with AI — leading with their ideas and letting technology handle the steps.

Start this week: Pick one type of content you write regularly and analyse three to five examples of your best work in that format. What patterns do you notice in your structure, tone, and approach? These insights become the foundation of your custom prompts.

Over the next month: Create a simple prompt template that includes examples of your writing style and clear instructions for the type of content you want to produce. Start basic and iterate based on your results. Develop a consistent review process. Track what’s working — time saved, quality improvements, engagement rates.

AI writing tools will only get better from here. Learning to use them effectively now builds skills and systems you’ll need to stay competitive as a content creator.

But remember what won’t change: your unique perspective, your experiences, your insights, and your ability to connect with your specific audience. AI can help you express these more effectively, but it can’t replace them.

Your ideas matter. Your experiences have value. Your perspective is unique. AI just gives you better tools to share all of that with the world.