I tried writing a blog post with ChatGPT, Claude, and Gemini — here’s how they stacked up

AI-Generating Blog Post Ideas

Before I go any further, I want to be hyper-clear on one thing: AI cannot read your mind.

It doesn’t know much about you — your tone, style, or how you think. But, give it the right context, and it might quickly catch on that you live for witty headlines and like to show up as helpful and unapologetically direct.

That’s why the first step in my process is setting the stage for expectation, just like you’d do if you were onboarding a new team member.

So before I put these tools to the test, I gave them the same background information, and then fed them the same prompts.

Want to learn more about my approach? I recently wrote about writing on-brand AI content and shared my approach (and included tips from other pros).

Sample Prompts

Here’s the sample prompt I used to introduce myself to the new team members, err… tools.

“Hi [Tool Name]. I’m a marketing strategist and content expert. I write for an audience of business owners, marketers, and creative professionals who want to improve their visibility, connect more deeply with their audience, and use messaging as a lever for business growth.

I have expertise in email marketing, messaging strategy, and using AI tools like ChatGPT and Claude to improve workflows and copy. My style is direct, conversational, and strategic.

My content goals:

• Show my POV and build trust through actionable, lived-experience insights.
• Help my audience write and communicate more effectively using tools like AI.
• Spark conversations with people who value clear thinking and human-centered content.

About my audience:

• Coaches, consultants, service providers, and content creators.
• Many are testing or curious about AI, but not sure how to use it well.
• They value voice, connection, and strategic depth over cheap hacks.

Please let me know if you have any questions about me before I walk you through the next prompts.”

Once I had clarified everything, I gave each tool the same task:

“I’m looking for blog post ideas around AI-powered writing workflows and how to train tools to sound like you. Please generate 10 distinct, specific, and click-worthy ideas that avoid clichés. Include a short explanation for each.”

Of note — all three tools provided 10 great ideas, but my screen could only grab so many. So I am sharing screenshots of the smaller asks and a text list of the 10 post ideas.

How ChatGPT Performed

Oh ChatGPT, how do I love thee? Let me count the ways.

ChatGPT has a leg up here since it knows me (and, candidly, probably where I live, too.) I’d say all of these tools are solid.

Here are ChatGPT’s top 10 ideas:

1. The First 5 Prompts I Give Any AI Tool Before I Let It Touch My Content
2. How to Build an AI Writing Workflow That Doesn’t Flatten Your Voice
3. Training ChatGPT to Write Like You? Start with These 3 Voice Anchors
4. What AI Can’t Hear (Unless You Teach It): The Hidden Signals in Your Writing Style
5. From Blank Page to Final Draft: My End-to-End Workflow Using ChatGPT
6. Why Your AI Writing Sounds Off — And How to Fix It Fast
7. How I Use AI as a Creative Partner (Not a Crutch)
8. Teaching ChatGPT to ‘Think’ Like You: A Framework for Strategic Content Prompts
9. The Exact Voice Guide I Feed ChatGPT (And How You Can Make Your Own)
10. Your Voice, On Demand: How to Create a Reusable Prompt Library That Sounds Like You

I decided to move forward with “The First 5 Prompts I Give Any AI Tool Before I Let It Touch My Content.”

How Claude Performed

Claude is fast. Like mind-blowingly fast. And I love that I didn’t need to do a ton of prompting to help it understand what I wanted.

I wasn’t planning to because I want to test the raw outputs of each tool, but it was an absolute delight that I didn’t need to go down that path.

Here are Claude’s top 10 ideas:

1. The Fingerprint Method: How I Trained Claude to Write in My Voice (With Examples)
2. Beyond Templates: Building Dynamic AI Writing Systems That Evolve With Your Brand
3. The Conversation-to-Content Pipeline: Recording Client Calls to Feed Your AI Writing System
4. Ethical Voice Cloning: Where to Draw the Line When Training AI to Sound Like You or Your Team
5. The 80/20 Rule of AI Writing: Identifying Your ‘Signature Phrases’ That Make Content Unmistakably Yours
6. Voice Layering: The 3-Pass Method for Adding Personality to AI-Generated Content
7. From Friction to Flow: Building Custom AI Tools That Eliminate Your Specific Writing Pain Points
8. The Authenticity Paradox: How AI Can Actually Make Your Content More Human, Not Less
9. Content Ecosystem Mapping: Designing AI Workflows That Connect Across Multiple Platforms
10. Voice DNA: Creating a Practical Style Guide That Both Humans and AI Can Follow

I decided to move forward with “The Authenticity Paradox: How AI Can Actually Make Your Content More Human, Not Less.”

How Gemini Performed

Gemini was a bit of a dark horse since I’ve had less experience with it. But again, it’s incredibly fast and came up with some really decent ideas. I’d argue that they were some of the most creative, even if a bit formulaic.

1. Beyond the Bot: The “Ghost in the Machine” Method for Training AI to Embody Your Brand Voice.
2. Stop Asking AI to “Sound Like You”: The Strategic Shift to Voice Emulation for Authentic Connection.
3. From Generic to Genius: The Feedback Loops That Turn AI Writing Tools into Your Uncanny Content Doppelgänger.
4. The “Human-in-the-Loop” Advantage: Crafting AI-Powered Content That Doesn’t Sound Like a Robot Wrote It.
5. Unlock Your Inner Author: Using AI to Bypass the Blank Page and Amplify Your Unique Perspective.
6. The AI Whisperer’s Toolkit: Essential Prompts and Techniques for Voice-Infused Content Creation.
7. More Than Keywords: Training AI on Your Core Messaging Pillars for Consistent Brand Communication.
8. Beyond Editing: The Strategic Art of Rewriting AI-Generated Content to Truly Own Your Voice.
9. The Ethical AI Author: Balancing Efficiency with Authenticity in Your Content Workflow.
10. The Future of “You” in Content: How AI Will Evolve Personal Voice and Brand Identity.

I decided to move forward with, “The ‘Human-in-the-Loop’ Advantage: Crafting AI-Powered Content That Doesn’t Sound Like a Robot Wrote It.”

The winner?

Tough call. I think all of them did great here.

Here are a few things I noticed that stood out and felt more AI-ish to me:

• Gemini used words like “crafted” which is often a giveaway for me.
• Both Gemini and Claude created repetitive structures for the titles with a punchy idea, followed by a colon and a more in-depth explanation.

To be more official, I used the following questions as a rubric to see how each performed:

Did the tool reflect an accurate understanding of me and my audience?

• ChatGPT: Yes, but also it has an unfair advantage. Did that come into play? We’ll never know.
• Claude: Absolutely. All of these topics sound like something I’d be interested in writing or reading about.
• Gemini: I think so. They’re quite interesting and I love some of the ideas it came up with.

Did it grasp my tone, POV, and priorities?

• ChatGPT: Yes. But, while it came up with stuff I’d write #rightnow, it felt a bit more generic, which feels like an interesting observation to me.
• Claude: A solid yes here.
• Gemini: Also yes.

Were the ideas fresh, not cliché? Would I want to click?

• ChatGPT: Yes, but while I liked the structure more here, I like some of the ideas better in the others.
• Claude: Absolutely. Have to work on the structures, but that’s fixable as we roll.
• Gemini: Same answer as for Claude.

So who is the winner?

Honestly, too soon to tell. I will say I was incredibly impressed with the level of thought and layers of detail in Gemini’s ideas.

Outlining the Post

Once I had a list of strong post ideas, I picked one to move forward with from each tool

Now came the next test: How well could each tool turn that idea into a structured, human-friendly outline?

Well, let’s find out! I gave each tool the same prompt.

“Please create an outline for a roughly 800-word blog post titled: “[insert title]”. Use H2 and H3 structure, and include a one-line summary of what each section should cover. Keep it human-first and logical.”

Here’s how it worked (note: for the sake of brevity, I omitted the one-line summaries and focused on the sections and key points for each):

How ChatGPT Performed

Here is ChatGPT’s outline:

The First 5 Prompts I Give Any AI Tool Before I Let It Touch My Content

• AI can be a powerful writing partner — but only if you train it right from the start. Here’s how I set the stage with five foundational prompts that save me time, protect my voice, and ensure I stay in control.
• Why AI Needs Context — Not Just Commands

• The myth of the “perfect prompt”
• Think of it like onboarding a new team member

• Prompt #1: “Here’s who I am and who I write for…”

• Why this matters
• Example prompt

• Prompt #2: “Here’s how I want my writing to feel…”

• It’s not just voice—it’s vibe
• Example language to include

• Prompt #3: “Here’s what I believe and how I talk about it…”

• Consistency beats cleverness
• Voice anchors and phrasing patterns

• Prompt #4: “Here’s how I want the reader to feel and act…”

• Tone follows purpose
• Example call-to-action framing

• Prompt #5: “Here’s the format I want you to follow…”

• The guardrails that keep things on track
• Reusable prompt template

• Final Thoughts: Better Inputs = Better Outputs

• A simple checklist for your next prompt session
• Start here, then evolve

Honestly, I think this is a pretty solid start. It feels pretty close to the vibe I want, and while yes, I need to share more of my opinions and writing style, I think we (me and GPT) could get to the finish line quite quickly.

Also, I love that it named the conclusion, whereas other tools just called it a boring ol’ “Conclusion.”

How Claude Performed

Here’s Claude’s outline:

The Authenticity Paradox: How AI Can Actually Make Your Content More Human, Not Less

• Challenge the common assumption that AI-generated content lacks humanity
• Set up the paradox: How the right AI approach can actually enhance authenticity
• Thesis: When used strategically, AI can free you to be more human in your content, not less

• The Real Problem Isn‘t AI — It’s How We’re Using It

• The Copy-Paste Trap
• The Strategic Alternative

• The Bandwidth Liberation Effect

• Automating the Mundane
• Reinvesting in Deep Work

• Practical Techniques for More Human AI-Assisted Content

• The Voice-First Workflow
• The Pattern Amplification Method
• The Human-in-the-Loop Editorial Process

• Measuring True Content Authenticity

• Beyond Engagement Metrics
• The Conversation Test

Not gonna lie, I love this. It feels high-brow and well-thought-out. This is definitely something I can see myself writing. Might take a bit more time, but this is a solid article I could really get behind.

How Gemini Performed

Here is Gemini’s outline:

The “Human-in-the-Loop” Advantage: Crafting AI-Powered Content That Doesn’t Sound Like a Robot Wrote It

• Introduction: Briefly introduce the allure and the pitfalls of AI-generated content, emphasizing the crucial role of human oversight in maintaining authenticity.
• The Siren Song of Pure AI Generation (and Why It Falls Flat)

• The Promise of Effortless Content Creation
• The Unmistakable Taste of the Algorithm
• Why “Good Enough” Isn’t Actually Good for Your Brand

• The “Human-in-the-Loop” Framework: Your Secret Weapon

• Defining the “Human-in-the-Loop” Approach
• The Critical Stages for Human Intervention
• Why This Hybrid Approach Delivers Superior Results

• Practical Strategies for Injecting Your Humanity into AI-Generated Content

• Training AI with Your Voice: Beyond Basic Prompts
• The Art of Strategic Rewriting: Adding Nuance and Perspective
• Fact-Checking and Ethical Oversight: Maintaining Trust and Accuracy
• Iterative Refinement: The Power of Feedback Loops

• The Future is Hybrid: Embracing Collaboration with AI

• AI as a Powerful Assistant, Not a Replacement
• Building a Sustainable and Authentic Content Workflow

Short answer — I don’t love it. It has some great ideas, but it feels less tactical and substantive. More importantly, I think there’s a pretty big lift to get it into my voice and share my thoughts.

The Winner?

Before I go any further, there wasn’t a clear winner, but there was one I didn’t like as much, and you’ve probably guessed it. Sorry, Gemini. Better luck next time.

That said, here are the questions I considered:

Did the outline flow logically?

• ChatGPT: Yes!
• Claude: Yes.
• Gemini: Yes.

Were the section headers helpful and clear?

• ChatGPT: Yes.
• Claude: Yes.
• Gemini: Yes.

Was there enough depth/detail for each section?

• ChatGPT: Absolutely.
• Claude: Absolutely.
• Gemini: This one is a no. Instead of offering specific advice, it offered a higher-level generic observation that I’d need more help thinking through.

Did it support the kind of blog post I’d want to write?

• ChatGPT: For sure.
• Claude: Absolutely. 1 million percent.
• Gemini: Honestly, not really.

Based on my gut-check responses to these questions, I’m taking back my statement about not having a clear winner. There was one — and it was Claude.

Writing the Post

This is where it gets interesting.

Coming up with ideas and outlines is fun, but writing the thing? That’s the real wild card here. And I had an inkling of what to expect — a hypothesis, if you will.

Each platform already had an overview of who I am and who I serve, but not how I write. So I shared this prompt with each platform:

“Here are three samples of my writing style so you can get an idea of how I think and what I believe. Please internalize this tone and keep it consistent as you help with idea generation, outlining, and writing.

Please let me know of any questions before I give you the next prompt.”

Each came back with a serviceable description of my writing style, layered in with what I shared in the very first prompt.

Next, I shared this prompt:

“Based on the outline you just created for a blog post about writing with AI, write a post. Keep it direct, helpful, conversational — and please, for the love of clarity, be specific when you share examples — leaning into thoughts you’ve identified in my examples, and avoid any cliche AI constructions or phrasing.”

Here are the first paragraphs that each platform came up with. For the sake of brevity, I won’t share the full article each developed. However, this should give you a sense of content, approach, and voice.

How ChatGPT Performed

Initial thoughts?

This sounds like me. But, there are definitely some AI-sentence structures that I try to keep to a minimum — namely:

• Question? Followed by a disarming sentence and one clarifying the answer.
• Parallel construction. Don’t just do this … Do this instead.
• Three short sentences starting with the same word.

Now, I’ve been known to do ALL of these. But I always watch for content like this because it just feels, well, average.

How Claude Performed

Okay — sort of a swoon here, because even though this feels more AI-like than most of what I’d write, it also is sort of close to something I’d write.

Here’s where it falls short. I’d never:

• Open a post with “Look, I get it.” I’d save that for later in the content.
• Talk like “Every time someone would rave…”
• Make weird-as-all-get-out analogies like “a committee of robots trying desperately to pass the Turing test.”

BUT, I don’t think it would take that much work to polish it up and get it into something I’d say.

How Gemini Performed

I’m going to stick to my guns here. I still absolutely love the “Human-in-the-Loop” idea. It got an A from me.

Execution is a flat C-minus. Sorry, Gemini, I realize this is a first date and all, but, this definitely doesn’t work for me.

Despite the title promise of not sounding like a robot wrote it, this definitely screams AI. Here are the tells for me:

• The word “crafting”
• Bringing in mid-conversation “We’ve all seen…”
• Language that sounds good, but realistically feels empty.
• Analogies that are catchy, but again, feel unsubstantial “The Siren Song.”

This isn’t goodbye, though, Gemini. We’ll still edit, and after that, I’m willing to give you another chance.

The winner?

It’s a close one, but Claude is in the lead, followed CLOSELY by ChatGPT. Gemini is nowhere near the finish line.

Here are the metrics I considered:

Did the draft feel like me?

• ChatGPT: Yes!
• Claude: Yes!
• Gemini: So. Much. No.

Was it structured well, with logical flow?

• ChatGPT: Yes, I’ll give it that.
• Claude: Yes!
• Gemini: I’ll give this one a yes. Flow was not the issue, the substance was.

Was it too robotic, or did it show personality?

• ChatGPT: A bit on the robotic side, but close and an easy fix.
• Claude: Same answer as ChatGPT.
• Gemini: It was so robotic it almost circled around and reinvented robots. (That means it was bad.)

How much editing would I need to do to make this publishable?

• ChatGPT: Pretty light. Just need to remind it what not to do, and layer in a few personal examples.
• Claude: Same answer as for ChatGPT.
• Gemini: This isn’t even worth pursuing for me.

For all of these, if I was really tackling a serious post, I’d probably have jotted down notes, or shared a voice recording or transcript with my two cents between Idea and Outline stages.

But, without that context, and just with what I fed the tools about me, ChatGPT and Claude both did a solid job.

Editing The Post

In all fairness, AI first drafts almost never hit the nail on the head the first time. Even if an AI-generated blog post is spot on for structure or tone, a human touch is needed. I’ve found it to be the difference between “Meh, that’s okay, I guess” and “Cue the confetti, that’s awesome!”

Something you’ll hear me say a lot is that mediocre is no longer good enough. It never really was, but a lot of mediocre content passed the sniff test before AI.

These days, your stuff has to be good — quality over quantity and all that. (Though, AI does help you create more content, faster if you use it right.)

If I was working on each of these posts, there is probably a lot I’d do to get more specific, including:

• High-level feedback.
• Taking things section by section.
• Reviewing the whole piece for consistency and clarity.
• Taking a final proofread spin.

However, to see how these tools responded to the same prompts, I once again gave them the same prompts. Namely, I focused on my personal brand of robotic red flags. I shared partial lists in feedback for each tool, and compiled them together here.

With that in mind, my goal here wasn’t to clean up grammar. AI usually does a decent job of that. I wanted to see which could take my changes and run with them to AI-generated blog posts that sound like me — or at least a bit more.

Here’s the prompt:

“Please revise this draft to tighten flow, improve rhythm, and make it sound more natural. Prioritize clarity and tone consistency. Do not add fluff.

Avoid the following:

• The word crafting
• Starting with mid-conversation phrases like “We’ve all seen”
• Language that sounds nice but says nothing
• Analogies that are clever but empty, like The Siren Song
• Openers like Look, I get it
• Phrases like Every time someone would rave
• Weird analogies like a committee of robots trying to pass the Turing test
• The question > softening > answer structure
• Parallel constructions like Don’t just do X. Do Y
• Three short sentences that start the same way
• Fluffy generalizations — make it specific and grounded”

Here are before and after screenshots to help you see the changes each made.

How ChatGPT Performed

I realize these screenshots are quite small, but to see the differences in sentence structure, I think it’s helpful to see the side-by-side edits.

Here’s the screenshot of the after version:

I’d give ChatGPT an A for tightening up and removing the constructions I don’t like. Sounds like me. Plus, it’s tight and to the point … and it just might make a great LinkedIn carousel or post.

How Claude Performed

I got chills! Claude was able to take my feedback and do a great job implementing it.

Now, here’s a larger version so you can read word for word:

Is it cheating on ChatGPT if I say #obsessed? Because Claude has really nailed this. Is it perfect? No, but is it really close and sounds much more like me? Yes.

How Gemini Performed

Here’s the side-by-side view:

And the single view:

I admit, after the fluff from the last round, I did NOT have high hopes here.

And I was pleasantly surprised. Does it sound like me? Not quite, but Gemini is getting closer. More importantly, it took the feedback I gave it and ran with it. Still needs more substance. But, a huge chunk of the fluff I hated is gone.

Based on this, Gemini is back in the running as a tool I’d like to play with some more. I think we could probably get there.

Overall Best Blog Post Generator?

Oh boy. This is hard.

I still am head-over-heels for ChatGPT. It’s my go-to for everything from finding kid-friendly recipes to planning my garden (Zone 5B, in the house!), and of course, AI-enabled workflows.

However, I don’t think ChatGPT performed the best within the constraints of this test. Off the top of my head, I wonder if we’re too close. ChatGPT knows the direction I’m likely to go and it often goes there, focusing on short and pithy over specific and descriptive.

Can Claude “learn me,” too? Maybe. Is Claude going to win every test? Of course not.

But Claude did the best when it came to the limited information I provided:

• Coming up with creative blog post ideas.
• Developing a solid, descriptive (aka not generic) outline.
• Fleshing out a well-structured and specific article.
• Editing to my heart’s content.

I also know that if I were to do more rounds of edits, we could absolutely get there. And, while Gemini didn’t win this round, I wouldn’t count it out from my next round of AI-generated blog posts or tests. I think it has some solid potential!

Where These Tools Can’t Replace Humans

Simply put, by its very nature, AI cannot do the one thing humans universally excel at — be human.

AI is, however, a phenomenal assistant and growing more impressive by the day. The biggest drawback (which is also good in a “will robots take over the world?” level of stress) is that it can’t think like me or think for me.

I have been able to use AI platforms to poke holes in my ideas, but as inherent people pleasers, I’ve found you must specifically request constructive criticism.

Otherwise, they are great at mirroring your cadence and emulating your structure. They’re just missing the you-factor of your lived experience and perspective.

The Real Magic Is in the Middle

I went into this thinking ChatGPT would be the winner. After all, it’s the tool I use daily, the one that’s already seen how I write, what I think, and how I work.

But Claude surprised me in the best way. It took direction, offered creative ideas I hadn’t considered, and turned around edits that actually sounded like me. Gemini even redeemed itself by the final round, which makes me curious to play more with it, too.

At the end of the day, I’m of the opinion that the best tool for AI-generated blog posts isn’t the one with the most features. It’s the one that helps you write faster, think better, and sound more like you. And that means it might change from time to time, or even from article to article.

And if this post convinced you to finally give AI a real shot — or refine how you’re already using it — then great. Because writing great content with AI isn’t about replacing yourself. It’s about extending your voice.

Editor’s note: This post was originally published in October 2023 and has been updated for comprehensiveness.