In 2024, ChatGPT started adding links to answers. This led to a frenzy of investments and startup activity, spawning a whole industry of "AI visibility" tools. VCs and marketing pundits now claim that search is dead and AI will take over. New acronyms like AEO (AI Answers Optimization), GEO (Generative Answers Optimization), and LLMO (LLM Optimization) begin to outshine the tired SEO. Or at least update it as AI SEO. Is it just hype or is there more to it? This post summarizes my months of research to get you up to speed on this emerging marketing channel.
What is AI visibility?
Put simply, AI visibility is about whether a brand or product is seen in AI answers.
By AI, we really mean LLM-based assistants that are used by people:
ChatGPT — used by everyone
Claude — used by many developers through Cursor
Google's Gemini — shows up in AI Overviews
Meta AI — Facebook's product
Grok — Elon Musk's product
DeepSeek, etc.
Each of these LLMs has several models, and new versions are actively developed.
"AI" could also include specific apps like Perplexity that use different models.
AI answers are generated based on 2 main inputs:
Training data of the model, usually the whole internet. This can be outdated by several months, and models rarely disclose cut-off dates.
RAG (Retrieval-Augmented Generation), which gives LLM access to external data not included in their training. This can include access to knowledge bases, but in practice, it's usually web search results.
For search results, each AI assistant uses different search engines:
ChatGPT uses Microsoft's Bing
Claude uses Brave
Gemini uses Google (obviously)
The way web search tools work for each AI assistant is also likely different, depending on what they prioritize (performance, comprehensiveness, token budget, etc).
As a result, AI visibility is fragmented across many different LLM models, search engines, and other tools.
With more acronyms, things don’t get easier.
AEO, GEO, LLMO, AI SEO: Optimizing AI answers for brands and products
AEO and other acronyms mean slightly different things, but they are all about AI visibility optimization.
AEO is Answer Engine Optimization — specifically about showing up in answers,
GEO is Generative Engine Optimization — showing up in generative responses, i.e. answers
LLMO is Large Language Model Optimization — showing up in LLM output, implies a focus on the output based on the training data and not augmented with search results.
AI SEO — the “SEO” or oprtimization for AI answers.
In practice, these acronyms are about the same thing: getting into answers that AIs show to your buyers.
The reason we have so many acronyms is that it's a new industry. There is still no consensus about how to call it.
But even if AEO is new, it's closely related to the oldest digital marketing discipline: SEO.
AI Engines Optimization vs SEO
As we've seen, AI answers are generated based on:
Training data (the whole internet, outdated),
Augmented data from search results (new info).
To answer your questions about crypto wallet seed phrases, AI can just rely on its training data. A lot of content was written about it over the years and the basic don’t change.
But to answer your prompts about "best crypto wallets in 2025", AI assistant will search the web for you. It will read the results, summarize them, and give you a confident answer.
So SEO is fundamental to AEO.
To be visible to AI, you need to be first visible in search. It’s really about AI SEO.
As a result, the same things that are important for SEO are also important for AEO or AI SEO:
Keywords
Backlinks
Internal links
But most crucially for AI SEO, keywords are derived from prompts. And while there is a limited number of keywords (e.g. "crypto wallet"), there is an infinite variety of related prompts. The way a prompt is structured will also cause the AI assistant to search for different keywords.
The are more differences between traditional SEO and AI SEO:
AI assistants read top search results and can go beyond the top-10
AI assistants launch multiple searches simultaneously
AI assistants don’t care about keywords and focus on semantic relevancy
A lot more stuff is also happening after AI assistants access web pages:
They will look for relevant text chunks
They will assess the authoritativeness
They will check the proofs and sources. In practice, they read like an educated human would, but without getting swayed by emotions, clickbait, or flashy images.
Yeah, AI assistants also currently ignore most images, video, and any interactive JavaScript content.
For now, AI assistants prefer boring old conversational text, written in full paragraphs.
How to measure AI visibility
Several metrics can be used to measure AI visibility of a brand or a product:
% of answers to target prompts where you're mentioned — absolute and relative to competitors,
Position or rank in answers vs competitors,
Sentiment in answers,
Clicks to your website from AI assistants.
A wide range of AI visibility tracking tools emerged to help with this.
Unlike traditioanl SEO, AI visibility doesn't yet have dominant and easy-to-use tools like SEMRush or Ahrefs where you can plug your website and track rankings.
Instead of keywords, AI visibility is based on prompts. The general approach is straightforward:
Define your prompts
Query AI assistants with these prompts
Record their answers and analyze them
But there is a lot of complexity:
Which prompts to use and how many?
How often to query? Monthly, weekly, daily?
How to analyze the answers?
Every AI visibility tracking tool answers these questions differently. I've tried dozens of tools, and that's a subject for a separate post. Some tools generate prompts for you, while with others, you need to create them manually. Some tools query daily, in others, you choose when to do it.
Rather than obsessing over tools, it’s more important to understand the high level approach to AI SEO.
How to Optimize AI Visibility
The best way to improve AI visibility for your product is to carefully observe what AI cites now reverse engineer that.
Track sources that are cited: your website, competitors, third-party websites
If your own website is cited, create more content, covering additional topics in a similar style and format
If your competitors are cited, create content similar to theirs
If third-party websites are cited, get mentioned in specific URLs that are already being cited
Keep in mind that not all products are equal. Older, bigger brands have a lot of history and presence in LLMs' training data. You should be more interested in smaller, younger brands that frequently show up in AI answers, punching well above their weight.
As you can see, the approach is straightforward. But it is a lot of work. There is no magic SaaS that will make you "rank in AI" with a click of a button.
But that's great. That's a major opportunity: few understand how this works, and even fewer will do the work.
Get ahead now
The AI answers optimization is where SEO was in the early days.
The competition is relatively low. The systems aren't yet as sophisticated. And building a solid foundation now will set you up for success in the near term and in the more competitive future.
It's easier to learn this together. Subscribe now to learn what works.