Recently I joined the SEO in 2023 podcast series with David Bain — part of a Majestic series of additional insights — where we broke down something I call the hybrid model for AI-assisted SEO. The conversation focused on a challenge that every SEO was grappling with at the time and is still relevant today: how to use AI correctly, rather than blindly, so that the output is actually good. Below are the key insights from that session turned into a practical guide.
Watch the video: SEO in 2023 — Victoria Olsina on How to Use AI for SEO
What Is the Hybrid Model for AI SEO?
The hybrid model is a simple idea: human strategy combined with AI execution, with human review at the end. Neither fully automated nor fully manual — the best of both.
The workflow looks like this:
- The human defines the strategy — seed keywords, target audience, content angle, point of view
- A well-constructed prompt passes that strategy to the AI
- The AI produces a draft
- The human refines, fact-checks, and publishes
What makes this different from just using ChatGPT? The prompt quality and the human judgement at both ends. A weak prompt produces generic output regardless of how powerful the model is. A strong prompt — one that specifies the audience, the format, the role, and includes examples — produces output that is far closer to usable on the first pass.
Why the Human Layer Cannot Be Removed
AI hallucinates. This is not a minor caveat — it is a fundamental characteristic of how large language models work. They produce the most statistically likely continuation of a prompt, not necessarily the factually correct one. For regulated industries, technical content, or anything where a factual error has real consequences, a human review step is non-negotiable.
The fix is not to avoid AI. It is to build fact-checking and expert review into the workflow as a permanent stage, not an afterthought.
How to Use AI for SEO Keyword Research
Keyword research is one of the clearest use cases for the hybrid model. The strategic layer — deciding which seed keywords to pursue, understanding the competitive context, making judgement calls on intent — stays with the human. The mechanical layer — clustering, grouping, intent classification — moves to AI.
AI for Topic Clustering
A simple, effective prompt for keyword clustering:
“Act like an SEO specialist and give me semantically related groups of keywords from this list.”
This one instruction reliably organises a flat keyword list into logical topic clusters, saving significant manual time. With GPT-4’s ability to accept CSV uploads, you can paste a full keyword export directly and get a structured cluster output in seconds.
AI for Intent Classification
AI is reasonably good at classifying keyword intent — informational, commercial, transactional — but it is not infallible. Tools like Semrush use machine learning for intent classification too, and experienced SEOs will sometimes disagree with their outputs. The right approach: use AI to classify at scale, then apply human judgement to the cases that look wrong.
Using AI to Generate Title Suggestions
Once intent is established, AI can suggest blog post titles quickly. A useful prompt structure:
“This is a blog post with informational intent targeting [keyword]. Suggest five title options.”
Use these as a starting point, not a final answer. A native English speaker or experienced copywriter will often refine the winning title further.
Producing AI-Assisted SEO Content That Sounds Human
Generic prompts produce generic content. The way to get AI output that sounds like a specific brand or person is to give it examples rather than descriptions.
Training the AI on Client Style
For product and service pages especially, brand voice matters. The method: collect three to five examples of existing content the client considers high quality, paste them into the prompt, and instruct the AI to match the style. The model infers tone, register, sentence length, and vocabulary from the examples far more reliably than from written descriptions like “professional but approachable.”
For clients with larger content libraries, even a small sample of three well-chosen examples makes a significant difference to output consistency.
What AI Cannot Do Yet
AI is capable of producing serviceable first drafts for most structured content types. What it still struggles with is genuine originality — surprising punchlines, unexpected angles, content that reflects a specific human experience.
As a stand-up comedian, this is something I notice directly. AI can generate jokes. It cannot reliably land a punchline. It can start something funny but cannot always finish it. For content that needs to be genuinely distinctive — not just competent — the human voice still needs to lead.
Using Perplexity and Google Scholar to Fact-Check and Enrich
One of the most practical tools in the hybrid workflow is Perplexity.ai for research. Unlike standard AI chat tools, Perplexity cites its sources. It surfaces academic papers, scientific research, and reputable references for a given topic, which can then be verified and added to the content as external links.
Google Scholar serves a similar purpose for more rigorous research requirements. For nutrition content, compliance documentation, technical B2B content, or any area where citing a credible source is important, both tools reduce the risk of publishing unverified claims.
The Bigger Picture: Strategic Thinking Is What AI Cannot Replace
The most important career advice for SEO professionals in any era of AI is the same: develop strategic thinking, not just execution skills.
AI can write a blog post. It cannot decide whether that blog post should exist, where it fits in a content funnel, how it connects to a paid campaign, or what it says about the brand’s positioning. Those decisions require context, experience, and judgement — none of which the model has.
The SEOs and content professionals who will continue to build strong careers are those who can zoom out: who see how a single blog post connects to video, to social, to email, to the broader strategy. The work of repurposing, distributing, and managing content across channels is increasingly automated — which raises the value of the person who can design the system rather than just execute within it.
Conclusion
The hybrid model is not a compromise between AI and human effort. It is the correct way to use AI for SEO: strategy and judgement from the human, scale and speed from the machine, quality control from the human again at the end.
Start with a clear prompt. Include examples. Let the AI draft. Review everything before it publishes. And keep updating your workflow — in this space, a week of not paying attention is genuinely expensive.
Here’s every single interview of the Majestic podcast series:
- SEO in 2023: The importance of using hybrid models for AI
- SEO in 2024: Build your own prompt libraries for your clients
- SEO in 2025: Improve the output from AI by creating custom GPTs for your clients
- SEO in 2026: Customise AI to automate your SEO processes
Frequently Asked Questions
What is the hybrid model for AI SEO content creation?
The hybrid model is a workflow in which human strategy and AI execution are combined, with human fact-checking and review at the end. The human defines the seed keywords, audience, content angle, and prompt structure. The AI produces a first draft. A human then refines, verifies facts, and publishes. This approach produces faster, more consistent output than fully manual content creation, while avoiding the factual errors and generic tone that come from over-relying on AI without human oversight. The hybrid model works because it applies each input where it has the most value.
How do you use AI for SEO keyword research without losing strategic control?
Use AI for the mechanical stages — clustering semantically related keywords, classifying intent at scale, generating title suggestions — while keeping the strategic decisions with the human. Seed keyword selection, competitive context, and final intent judgements should remain human-led, because AI classification tools (including Semrush’s own machine learning) are not always accurate. A practical starting point: export your keyword list, upload it to ChatGPT, and prompt it to “act like an SEO specialist and give me semantically related groups of keywords.” Review the clusters and adjust based on your own understanding of the market.
How do you get AI to write in a client’s brand voice?
Feed it examples rather than descriptions. Collect three to five pieces of existing content the client considers high quality — their own writing or competitor content they admire — and paste them directly into the prompt. The AI infers tone, register, sentence length, and vocabulary from real examples far more accurately than from written instructions like “professional but approachable.” For product and service pages where brand voice matters most, this examples-first method produces noticeably more consistent output than generic prompting.
What are the main risks of using AI for SEO content?
The two primary risks are hallucination and generic output. AI models produce statistically likely continuations of prompts, not factually verified statements — which means they can and do generate incorrect information confidently. For regulated industries, technical B2B content, or any content where factual accuracy matters, human fact-checking is non-negotiable at every stage. Generic output is a separate problem caused by weak prompts. Providing specific role instructions, audience context, format requirements, and real examples in the prompt significantly reduces how generic the output is.
What SEO tasks should remain human-led even with strong AI tools available?
Strategic decisions should remain human-led: which keywords to pursue, how content fits into the funnel, how pieces connect across channels, what the brand’s positioning requires. These decisions require context and judgement that AI does not have. Execution tasks — drafting, clustering, intent classification, title generation — are strong AI use cases. The distinction matters because misapplying AI to strategic decisions (letting it choose the content strategy) produces directionless output, while misapplying humans to execution tasks (manually clustering thousands of keywords) wastes expertise.
How does the hybrid AI model affect SEO careers and hiring?
Senior SEO skills — strategy, analysis, cross-channel thinking — become more valuable as AI handles more execution work. Junior roles that consisted primarily of producing first drafts, formatting content, or running repetitive research tasks are the most exposed to displacement. The practical career advice: develop strategic thinking as early as possible. Learn how a single piece of content connects to a broader campaign. Understand the funnel, the distribution channels, and the measurement framework. SEO professionals who can design content systems rather than just execute within them will continue to be in demand regardless of how capable AI tools become.
Need help turning ideas like this into working systems?
This is the kind of AI marketing work we do with teams that want repeatable workflows, not one-off experiments. Let’s build yours →
