Everyone’s talking about AI content writing. Most of them are doing it wrong. They’re using AI to replace their voice instead of amplify it. After 20 years of writing content that actually ranks, I’ve found a better approach — using tools like Perplexity and Claude not to write for me, but to sharpen my thinking, nail my angle, and find SEO opportunities I’d miss on my own.
Here’s the framework I use, complete with the actual prompts. Steal them.
The Problem with AI Content Writing
Let me be blunt. Most AI-generated content is garbage. It’s technically correct, grammatically clean, and completely soulless. Google knows it. Your readers know it. And if you’re building a personal brand or a business, it’s actively hurting you.
The issue isn’t AI itself — it’s how people use it. They type “write me a blog post about SEO” and publish whatever comes out. That’s not content strategy. That’s a slot machine.
What actually works is using AI as a research partner and thinking tool while keeping your authentic voice front and center. That’s what I’m going to show you.
Step 1: Use Perplexity to Build a Research Foundation
Perplexity has become my go-to for content research. Unlike a regular search engine, it synthesizes information from multiple sources and gives you citations. It’s like having a research assistant who actually reads the articles instead of just listing them.
Before I write anything, I run research prompts to understand what’s already out there and where the gaps are.
Perplexity Research Prompts
Here are the exact prompts I use to build my research foundation:
Prompt: Gap Analysis
“What are the top 10 articles ranking for [your keyword]? Summarize their main arguments, identify what they all agree on, and tell me what none of them are covering.”
Prompt: Data Points
“Find recent statistics and studies (2024-2026) about [your topic]. I need specific numbers with sources I can link to.”
Prompt: Counter-Arguments
“What are the strongest arguments against [your position]? What would a skeptic say? Include real examples.”
This gives me a solid foundation of facts, data, and competing viewpoints — all before I write a single word of my own.
Step 2: Use Claude to Cement Your Unique Angle
This is where it gets good. Most people use AI to write content. I use it to think through my angle before I write. Claude is exceptional at helping you find the gap between what everyone else is saying and what only you can say.
Claude Prompts for Finding Your Angle
Prompt: Angle Discovery
“I’m writing about [topic]. Here’s what every other article says: [paste your Perplexity research]. My background is [your experience]. What unique angle could I take that no one else is covering? Give me 5 options with a one-sentence hook for each.”
Prompt: Authenticity Check
“Here’s my draft outline for a post about [topic]: [paste outline]. Read this critically. Where am I saying something generic that anyone could say? Flag every point where I should add a personal story, specific example, or original opinion instead.”
Prompt: Experience Mining
“I have [X years] of experience in [your field]. I’m writing about [topic]. Interview me — ask me 10 questions that would pull out stories, lessons, and opinions only someone with my experience would have. Things I might not think to include on my own.”
That last one is my favorite. Claude essentially interviews you, and your answers become the raw material for the most authentic parts of your post. No AI could generate those stories — they’re yours.
Step 3: Optimize for SEO Without Losing Your Voice
Here’s where most people go wrong. They write something authentic, then run it through an AI “optimizer” that strips out all the personality. Don’t do that.
Instead, use AI to find SEO opportunities you can weave into your existing draft naturally.
SEO Enhancement Prompts
Prompt: Semantic Keyword Mapping
“My primary keyword is [keyword]. List 20 semantically related terms and questions people also search for. Group them into clusters. For each cluster, suggest where in my article it would naturally fit without forcing it.”
Prompt: FAQ Generation for Featured Snippets
“Based on this article about [topic], what are the 5 most likely questions someone would ask Google that this content could answer? Write the questions as people would actually type them, not as formal queries.”
Prompt: Title and Meta Optimization
“Here’s my blog post title: [title]. Give me 10 alternative titles that include my target keyword [keyword], are under 60 characters, and still sound like a human wrote them — not an SEO robot. Rank them by click-worthiness.”
The key is that you write the content first, then use AI to identify where you can strengthen the SEO without compromising what makes it yours. As I covered in my post about what actually works in SEO in 2026, search engines reward genuine expertise — not keyword games.
The Complete Workflow
Here’s my actual process from idea to published post:
- Research with Perplexity — Understand the landscape, find gaps, gather data
- Find your angle with Claude — Use the interview prompt to mine your real experience
- Write the draft yourself — This is non-negotiable. Your voice, your stories, your opinions
- Enhance with Claude — Run the authenticity check to flag generic sections, then replace them with personal examples
- Optimize for SEO — Use semantic keyword mapping to find natural opportunities
- Final read-through — Read it out loud. If it sounds like AI wrote it, rewrite those sections
The whole process takes me about 90 minutes per post. That’s longer than just letting AI spit something out, but the difference in quality — and in how it ranks — is massive.
Why This Approach Wins
Google’s Helpful Content guidelines are clear: they reward content that demonstrates real experience and expertise. Their stance on AI content isn’t anti-AI — it’s anti-low-quality. Content created with AI assistance that still demonstrates genuine expertise ranks just fine.
The E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) practically demands authenticity. A post stuffed with AI-generated filler fails the Experience test every time. A post where AI helped you organize and sharpen your real expertise? That passes with flying colors.
My approach to using AI in marketing has always been the same — let AI handle the heavy lifting on research and optimization while keeping the human element where it matters most.
The Bottom Line
AI isn’t going to write your best content. You are. But AI can help you research faster, think deeper, find angles you’d miss, and optimize for search without selling your soul. The prompts above are my actual toolkit — take them, adapt them, make them yours.
The writers who will win in 2026 aren’t the ones who produce the most AI content. They’re the ones who use AI to make their authentic content even better.
Frequently Asked Questions
Does Google penalize AI-written content?
Not specifically. Google penalizes low-quality content regardless of how it’s produced. AI-assisted content that demonstrates real expertise and provides genuine value ranks just fine. The problem is fully AI-generated content that lacks originality, personal experience, and depth.
Can I use these prompts with ChatGPT instead of Claude?
Absolutely. These prompts work with any capable LLM. I prefer Claude for the angle-finding and interview prompts because I find it gives more nuanced, less formulaic responses — but your mileage may vary. Use whatever tool you think better.
How do I know if my content sounds too much like AI?
Read it out loud. If you hit phrases like “in today’s digital landscape,” “it’s important to note,” or “let’s dive in” — those are AI tells. Real people don’t talk like that. Your content should sound like you explaining something to a colleague over lunch.
Is this approach worth the extra time versus just using AI to write everything?
If you’re building a brand or a business, yes. Fully AI content is a race to the bottom — everyone has access to the same tools. Your experience, opinions, and stories are the only things competitors can’t replicate. That’s your moat.
