How I Use AI to 10× My SEO Workflow (Without Losing My Voice)

I was wasting 6 hours a week on keyword research. Not because it's hard — because it's repetitive. The same process, same tabs, same copy-pasting. Until I built a system that lets AI handle the mechanical parts while I handle the thinking.

This isn't about replacing your brain. It's about giving your brain better inputs.

The Problem With How Most Marketers Use AI

The typical workflow looks like this: "Write me a blog post about keyword research." Then they publish whatever comes out. The result is 800 words of generic advice that reads like every other article on the internet.

Google's Helpful Content Update is specifically designed to punish this. And frankly, so does your audience — they can feel the difference between content that was thought through versus content that was generated.

The right use of AI is as a research and formatting layer, not a creative layer. Here's what that looks like in practice.

Step 1 — Keyword Clustering at Speed

I used to cluster keywords manually in Google Sheets. It took 2–3 hours for a proper 200-keyword list. Now I do it in 15 minutes.

My prompt:

Here is a list of 200 keywords from SEMrush for [topic].
Group them into clusters by search intent:
- Informational
- Commercial Investigation
- Transactional
For each cluster, suggest one primary keyword and list supporting keywords.
Output as a table.

The output isn't perfect. I always review and merge clusters that are too similar. But the mechanical work is done. I'm now just editing, not building from scratch.

"The goal is to compress the boring parts of research so you can spend more time on the interesting decisions — what to prioritise, what angle to take, what competitors are missing."

Step 2 — Competitor Gap Analysis

This used to take me a full afternoon. Now it takes 30 minutes. Here's the process:

I pull the top 5 ranking pages for my target keyword, copy their H2s and subheadings, and feed them into a prompt:

Here are the subheadings from the top 5 ranking pages for "[keyword]".
Identify:
1. Topics all 5 pages cover (table stakes)
2. Topics only 1-2 pages cover (differentiation opportunity)
3. Topics none of them cover (gap opportunity)
4. Questions likely to be in "People Also Ask"
Format as a structured outline.

What I get back is a content brief that's more thorough than anything I'd build manually. And more importantly — I can see exactly where I can differentiate instead of just writing "me too" content.

Step 3 — Draft Structure, Not Draft Content

Here's where most people go wrong. They ask AI to write the article. I ask it to give me an outline with reasoning.

Based on this content brief, suggest an H1, 5-7 H2s, and for each H2:
- What question it answers
- What evidence or example should support it
- Why it matters to someone searching "[keyword]"
Do not write the content. Just the skeleton.

Then I write the actual content. Every word is mine. The research is AI-accelerated. The voice is human.

The Full System in One View

Here's how these three steps fit into a weekly workflow:

• Monday: Pull keywords from SEMrush, cluster with AI (15 min)
• Tuesday: Pick target keywords, run competitor gap analysis (30 min)
• Wednesday: Build content brief skeleton with AI (20 min)
• Thursday–Friday: Write the actual content, add real examples, publish

Total AI-assisted time: ~65 minutes. Time saved vs. old process: ~5 hours. That's the 10× I'm talking about — not magical AI output, just less time on the mechanical parts.

What I Don't Use AI For

This matters as much as the above. I never use AI for:

• Writing any section of a published article
• Deciding what to prioritise (that requires business context AI doesn't have)
• Generating client reports (clients hired me, not a chatbot)
• Creating headlines (my best-performing titles are always written by me)

The goal is augmentation, not replacement. The moment a client reads your content and feels like they're reading a robot, you've lost the relationship.

That's the system. It's not flashy, but it compounds. Every hour I save on mechanical work is an hour I can spend on things that actually require a human.

— Suraj