AI Thinking

How to use AI to accelerate your thought leadership

...without destroying it

Katie Lips, CEO and Founder of Rebel Futures/16 June 2026
How to use AI to accelerate your thought leadership

76% linkedin users I surveyed say unedited AI content damages their view of the person posting it. The word most used to describe GenAI posts and articles on LinkedIn is “hollow”.

But what should you do about it if you want to create and share effective Thought Leadership, build a personal or business brand, and facilitate genuine and useful engagement?

I created a guide on how to use AI well in creating and publishing Thought Leadership content that people want to read. It’s for individuals, whether working independently or within a business or enterprise. People sharing their thoughts on LinkedIn are either using their own self-directed tools (they decided they wanted to try Chat GPT or Claude), or they are using tools mandated by their employer (Gemini or Copilot has been rolled out). This guide assumes you have some autonomy and control over how you use these tools.

Maybe you want to publish regularly on LinkedIn, but you don’t have time, or maybe you find writing hard, or maybe you want to improve the value of and engagement with your content, or maybe you have a team of people all publishing in different styles and you want to create a single, brand-led narrative. These are all reasonable things AI can help with.

AI is great in that it democratises content creation for the many, but it also then makes most content sound the same, unless you use it intelligently, know what it’s great at and what it’s not so great at, and put in a little more effort to get it to work for you and your business.

What the current problem with AISlop (the slew of well-intended, but obviously AI-written generic-sounding posts highlights is that more and more people want to get their message across, engage with communities and build profile. The problem with much of the AI-generated content out there, however, is that it doesn’t do these things.

AI-generated posts on LinkedIn often irritate their audience before they even get into the detail, so they won’t engage communities and won’t build a profile. The counter to this argument is that I know some people who’ve written or rather co-written posts with AI, and had huge engagement from them. But in these cases, they either already had a large following to begin with and/ or, the content of these posts is really interesting - a unique take on a topic, something completely new, something truly provocative. The point is that AI is not likely to write this sort of content for you by default, and needs coaching to do so.

The problem with AI-generated articles all seeming to be the same, with the same structure, flow, pace, urgency, style and format, is not because AI is bad at writing, but because it’s not been used well. If you tell AI to write something interesting, in your specific style, leveraging your unique experience, from your point of view, with new insight, the output will be interesting. If you tell it to write a LinkedIn post, it will just write a LinkedIn post - that sounds like all the other GenAI LinkedIn posts.

The problem stems from users missing the fundamentals of the writing process and skipping straight to ‘write post’, before researching, considering, shaping, testing and drafting. When writers sit down to write a book, story, article or even a social media post, they do not start typing at the beginning and finish at the end, and then hit submit.They do not begin with a title or headline, then write a good opening argument, then develop a convincing body and finish with an exciting wrap-up.

Instead, before writing, they think, test, research, balance, have a strong point of view, leverage evidence, and consider consequences. Then, when writing, they refine and refine, perhaps even rewriting until the argument makes sense, and it lands in their voice.

THE FRAMEWORK

Here’s the complete framework: five thinking stages before you write a word, a skills file that trains AI on your voice, a prompt engineering template you can copy and use, plus an editing process that keeps ensuring the work is yours.

AI Assistants

This framework works with existing AI assistants; Claude, ChatGPT, Gemini, Perplexity and others. The differences between these tools don’t matter all that much, but some are particularly good at certain tasks. Perplexity is great for research and fact-checking with its live web access.

For drafting, writing and holding context across a complex argument, Claude and ChatGPT both perform well. If you work in a Google environment, Gemini integrates cleanly with Docs and Gmail, and likewise Copilot is the natural choice if you work in a Microsoft Office environment (although Copilot is not as strong as some of the others in open-ended reasoning).

Your AI platform may well have been decided for you by your Enterprise IT and which AI Assistant you use matters less than your ability to go deep with it, and adapt it to your needs. The more you use any of these tools, the better your prompts become, and the better the outputs get.

Where AI helps in the writing process

AI is good at writing; it’s even better if prompted well. But AI is capable of more than just the creation of the final output. It is useful and viable at so many more points in the writing process than most people use it for.

AI Assistants are great across the writing process, from defining your voice, to thinking, to drafting, to editing. This guide shows you how to use AI across these 4 areas.

GET SETUP FOR SUCCESS

Before you start, it’s worth taking the time to set up your work space properly. It will make every future piece of content you write with AI faster and better.

Most of the major AI assistants now support some form of persistent workspace (a place where your context, instructions and reference materials live permanently, rather than having to be re-established every time you open a new conversation. Claude calls these Projects. ChatGPT has a similar capability through custom GPTs and memory. Gemini has Gems and Copilot supports persistent instructions within its enterprise configuration. If your tool supports this, use it. It is one of the most underused features in AI-assisted writing and one of the most valuable.

This is what separates people who use AI well from people who use it starting from zero each time.

Here’s how to structure your space:

Reference folder - Create a folder inside your workspace if the tool supports it, or alongside it in a cloud document if not, where you keep the material relevant to your writing. Include research, data points, previous articles, client conversations or observations you want to draw on, and even competitor content you want to respond to.

One document per article - For each piece you’re developing, keep a single working document that holds everything in one place: output from your thinking stage, your prompt, your draft, your edits and your final version.

Prompts file - Consider keeping a document of prompts that worked well. When a prompt produces output you’re genuinely pleased with, save it. Over time this becomes your personal prompt library which will make new work faster to start.

Your skills file - Paste your skills file (see below) into the persistent instructions for your workspace.

DEFINE YOUR VOICE

Before you try to write anything, it’s vital to define your voice. Either you already have a voice (based on years of existing content) or you will need to define one. Either way, you need to tell your AI Assistant what it is.

Create a skills file

A skills file is a document you build once that tells your AI assistant who you are, how you write, and what you want to produce. It’s hugely underused. Usually, when you see AI-generated content that sounds like ‘all other AI-generated content’, it’s because the author has no skills file, so the AI assistant has no context.

In some tools you build skills files once and they persists automatically; Claude’s Projects, ChatGPT’s Custom Instructions, and Gemini’s Gems allow you to store this context forever. If you’re using a tool without persistent memory, create a simple text file and paste this in at the start before you start writing a new article.

4 Steps to create a Skills File:

Know your voice - Include or upload real examples of your own writing. Tell your AI assistant:

“This is how I write. Match this register, this rhythm, and tone.”

If you are new to writing, think about how you want to come across. You can even suggest authors you like and want to emulate, but add your own unique perspective.

Define your audience - You don’t need to be super detailed, but the more detail you have, the better. You could start with something like:

“Senior business and technology leaders navigating AI adoption. They are smart, busy, and they like specifics.”

Define your positioning - Tell your AI assistant about your perspective. Again, be as detailed as you like. Mine might be something like:

“I write from 20 years of digital leadership experience. I am interested in the human and organisational consequences of AI, not only in the technology itself. I am direct, provocative, and never a cheerleader.”

Have fun with this, and make it your own.

Specify what you don’t want - This step is brilliant in stopping AI defaulting to its most common patterns - you know, the ones that make it really obvious it’s AI. Rule out things you don’t want. Here’s an example:

“Never start with a rhetorical question. Never use bullet points unless the content is genuinely list-like. Never use the phrases ‘in today’s fast-paced landscape’, ‘it’s more important than ever’, or ‘let’s dive in’. Never open with ‘As a [job title]...’ Never end with ‘What do you think? Drop a comment below.’”

And so on….

Share your skills file - Save this document and share it with your AI Assistant. In Claude you can tell it to save it as a skills file. Once done, you’ll notice the difference immediately.

The above two things are things you need to set up once. Sure, you will want to revisit them occasionally, but once done they’re useful to you every time.

EXPAND YOUR THINKING

Now it’s time to start thinking about your article. The thinking stage is where the real work happens. Skip this step and you’ll produce content, but it may not be worth reading.

Test your premise - Many people leap straight to drafting without refining an idea. This simple step lets you test whether this idea really holds or if it needs refining. It reduces the risk of commenters pulling your argument apart. Ask your AI to test your premise with a prompt like:

“My hypothesis is [x]. Is this true in reality, or have I oversimplified? Show me evidence for and against this hypothesis? Are there any holes in this argument?”

Research your topic - You don’t need to know everything about your topic, but you need to be informed. AI Assistants allow us to create articles about things we don’t have all that much experience in, and that’s a danger. It means we could end up publishing a post really knowing as much as an authority on the topic would be expected to.

So use AI to identify what you’re missing and treat it as a learning exercise.

“I’m writing about [topic] for [audience]. What should I know about this? What recent data, examples or developments are relevant that I should incorporate?”

Check everything in the response and definitely don’t include any stats without verifying them.

Identify the counterargument - This is a really valuable stage and usually skipped. Content with only one view, with no acknowledgement of the counter argument can feel a bit novice, so including a nod to the alternative point of view makes your content feel better thought through.

“Here is my argument: [argument]. What is the strongest possible counterargument a smart, sceptical reader would make? What could I be missing or oversimplifying? What would someone who disagrees with me say?”

If your argument can survive this process, great. If there are strong counter arguments, you will be able to address them as you write.

Make sure you have an original angle - AI content tends to cover the most obvious angles, because AI is trained on historical data. It’s unlikely to come up with something “new new”. Either you will have your own unique angle, or you can ask AI to identify what’s missing in current coverage - but again, your own unique perspective trumps everything else for authenticity.

“What is the most interesting or under developed version of this argument? What angle are most people writing about this topic missing? What would make a reader think ‘I hadn’t considered that’?”

Consider second-order implications - Your post should make a point, but what’s the “So What?”. What are the actually exciting, (or dangerous, urgent, etc) implications of this now and next?

“If my argument is correct, what are the second-order implications? What else follows from this? What are the likely unintended consequences of this? What does this mean for [your audience’s specific context]?”

Only once you’ve completed these 5 steps are you ready to start writing.

DRAFT WELL

The quality of any AI output is determined by the quality of the prompt. Much of what is termed AISlop was likely created with poor prompts. Creating a structured prompt helps you not only tell your AI Assistant what you want, but helps you structure your own thinking too.

This prompt template has ten components. Use as many as you can and draft a structured prompt.

ROLE: [Who should the AI assistant be in this conversation?]

AUDIENCE: [Who is this content for? What do they care about? What do they know?]

GOAL: [What should the reader think, feel or do after reading the content?]

ARGUMENT: [What is the specific point you want to make?]

CONTEXT: [Why are you writing this now? What prompted it?]

TONE: [How should it feel? Direct? Warm? Challenging? Conversational?]

FORMAT: [What is the length and structure?]

AVOID: [Are there any specific arguments, conclusions or framings to stay away from]

VOICE: [Refer to your skills file or writing examples]

REFERENCE: [Optionally, add or reference a piece of writing that represents the quality you’re aiming for]

Here’s an example.

ROLE: You are helping me write a LinkedIn article in my voice. You are a skilled editorial collaborator, so your job is to help me articulate my thinking clearly and compellingly, not to replace it.

AUDIENCE: Senior leaders in professional services firms who are smart, busy and sceptical of hype. They understand their industry and their clients’ challenges. They care about competitive positioning, client relationships and the future of their firms. They create thought leadership content within their teams and personally, and are concerned that their output needs to be high value and authentic.

GOAL: I want the reader to realise they have a gap in their AI assistant rollout, to feel a sense of urgency around fixing it, and to feel that my company can help them.

ARGUMENT: Many enterprises are either rolling out AI assistants or allowing them to be used as Shadow AI. But teams are not getting training on how to use them well, so individuals are using them to create low-value AISlop that is harming individual and organisational reputations. This isn’t a technology problem. It’s a leadership problem, and teams need help using AI well as part of a considered contextual AI change programme.

CONTEXT: I’ve just published a piece on how people feel when they see AI-generated content on LinkedIn. This is the next Enterprise AI adoption argument that follows from it. I want to offer value to readers so they can instantly start seeing better outcomes when using AI in content creation. From this, I want them to feel they can contact my firm to ask for help in developing an enterprise-wide AI change programme.

TONE: This should be direct and challenging but not preachy. I want to provoke without lecturing. I respect my reader’s intelligence. I do not need to explain AI to them, but I am calling out where technology rollout alone, without proper cultural adoption and understanding could do more harm than good.

FORMAT: 1000 - 1250 words. Prose, not bullet points. No rhetorical opening question. No listicle structure. Short paragraphs. One strong closing line that earns the ending rather than summarising what I’ve already said.

AVOID: Don’t suggest firms should not use AI or that AI content is bad. Rather that it needs to be used well in thought leadership. Don’t end with a generic call to action asking people to comment or get in touch.

VOICE: Refer to the skills file you already have.

REFINE

Don’t take your draft and hit publish. There will be things you can improve and refine.

Rewrite - Take the structure if it works, but replace the language with yours. Add the specific detail only you have (personal stories, client conversations etc). Cut anything that sounds like it came from a model trained on a gazillion LinkedIn posts. Some people rewrite more than others. If the idea of rewriting an entire article seems like too much (why use AI in the first place?), treat this as an editing task.

Sense Edit - In a final edit, ask your AI assistant to identify any potential issues. You could ask:

“Does my argument lose momentum?” “Which sentence or paragraph could be cut?” “Does the ending land?”

Read it aloud - Reading aloud is a great way to identify issues with any writing. If it sounds like you talking, it’s good. If it sounds like a press release, or like GenAI, it might be time to start again.

Spelling and Grammar proofing - If you have made edits, ask AI to proof read. You can tell AI not to write anything new, but just to highlight any issues. You can also use this step to ensure your text is correct - for example that you use British, or American English spellings and grammar and so on.

AI Check - Not everyone minds if it’s apparent their content was assisted by AI. Most people use AI tools in one way or another so it may not be an issue for you. But no one wants their ‘Thought Leadership’ to sound 100% generated. There are tools you can use to highlight where your content sounds AI-generated, and where it does not. We like Pangram for effective AI detection.

Use AI to get there faster. Don’t use it to get there without you.

Rebel Futures helps business and technology leaders use AI as a genuine competitive advantage. If you’re thinking about how AI should be shaping your thought leadership strategy, we’d like to talk.

Katie Lips is CEO and Founder of Rebel Futures. She helps organisations grasp the opportunities at technology inflection points, from the early web and pre-iPhone mobile to cloud-based SaaS and now AI. She has held senior roles at EY and Aimia, and was Head of Digital Policy at Which?. Katie holds a Harvard Kennedy School qualification in Leading in Artificial Intelligence and is a Founder Institute graduate.

Related: explore our work on AI Adoption.

base44
Edit with Base44