From SEO to AI Search: Jarno S’s Journey Into Making the Web More Understandable to AI
“Build for people. Structure for understanding. Measure what changes.”
Search is changing.
For years, businesses and website owners have focused on questions like: Where do we rank on Google? How much traffic are we getting? Which keywords are bringing visitors?
But as people increasingly turn to AI systems such as ChatGPT, Gemini, Perplexity and Copilot for answers, another question is becoming increasingly important:
What does AI say about your business — and is it accurate?
For Jarno S, founder of AEOvara in Finland, this question sits at the intersection of technical SEO, AI search, structured data, content architecture and measurement.
Jarno's path into this work has been practical rather than purely theoretical. He has built websites, worked with visual content and photography through Kuvaajankulma, and experienced how difficult it can be for a good business to become discoverable online.
SEO became a natural part of solving that problem.
Then AI search changed the question.
Instead of asking only whether a website can rank, Jarno became interested in whether AI systems could understand a business correctly, represent it accurately and recommend it for the right reasons.
That thinking led to AEOvara and his work around Answer Engine Optimization (AEO), Generative Engine Optimization (GEO), and Large Language Model Optimization (LLMO).
From Traditional Search to AI Search
Jarno describes several related concepts that are increasingly appearing in conversations about search.
Traditional SEO is mainly concerned with earning visibility in search-result pages.
AEO — Answer Engine Optimization — focuses on structuring content so that it can answer a question directly.
GEO — Generative Engine Optimization — focuses on how a brand is represented and recommended in generative search experiences.
And LLMO — Large Language Model Optimization — is about making a website, its entities and its information understandable to large language models.
These approaches overlap, but for Jarno the broader shift is more important than the terminology.
The goal moves from simply trying to “rank for a keyword” toward becoming a reliable answer.
That distinction becomes particularly important when the interface between a person and the web is no longer simply a list of links.
The Question That Changed Everything
Jarno became interested in AI search when he noticed that people were increasingly asking ChatGPT, Gemini and Perplexity questions that they previously asked Google.
That introduced a new problem.
It was no longer enough to ask:
“Where do we rank?”
There was another question:
“What does the AI say about us, and is it accurate?”
That question became central to his work.
Jarno describes his work as having two audiences: the people reading a website and the AI systems trying to understand it.
That idea has influenced how he approaches website architecture and content.
A page should work for a person who is in a hurry, while also giving a system enough clarity to identify the business, its services, evidence and expertise.
That means clear page purpose, direct answers, logical headings, consistent facts and strong internal connections.
AI Visibility Should Be Measured
One of Jarno's strongest themes is measurement.
AI search can sometimes appear mysterious. Ask an AI system a question, receive an answer, take a screenshot and it can be tempting to conclude that a business is either visible or invisible.
Jarno argues that this isn't enough.
AI outputs can vary depending on the platform, wording, location and time.
A more useful approach is to create a repeatable measurement process.
His recommendation is straightforward:
Start with 8–15 real customer questions.
Keep the wording stable.
Test those questions across the relevant AI platforms.
Then record the results.
The goal isn't to create a perfect score.
The goal is to establish a trustworthy baseline that makes changes visible.
Think of AI Visibility Like Regression Testing
Jarno compares an AI visibility benchmark to a software regression test suite.
The analogy is particularly useful for developers.
If the test changes every time you run it, it becomes difficult to determine whether the software changed or the test changed.
The same applies to AI search.
If the prompt changes every time, you cannot reliably determine whether a change in the result came from an improvement in the brand's visibility or simply from asking a different question.
Jarno recommends keeping a stable core set of prompts while adding new questions over time when necessary.
It turns AI visibility from a collection of interesting screenshots into something that can be observed and compared.
A Mention Isn't the Same as a Citation
Another distinction Jarno considers important is the difference between mentions and citations.
An AI system might mention a business in an answer.
But that mention could be vague, incorrect or simply part of a long list.
A citation provides a stronger signal because it connects a specific claim to a source.
Both matter, but they answer different questions.
Visibility is useful.
But trusted and accurate attribution is more valuable.
For Jarno, metrics should therefore be separated rather than combined into one simplistic number.
He focuses on areas such as:
- Visibility
- Mentions
- Citations
- Accuracy
- Meaningful or qualified actions
The distinction helps businesses understand not only whether AI systems are talking about them, but how they are representing them.
Structured Data Helps AI Understand the Pieces
Jarno's work also places significant emphasis on Schema.org structured data and entity clarity.
He doesn't describe Schema.org as a magic solution.
Instead, he sees it as a useful shared vocabulary that can help systems identify things such as:
- Products
- Services
- People
- Organisations
- Locations
- FAQs
- Relationships between entities
But structured data becomes much more useful when it matches the visible content on a website and the information remains consistent across the site.
Adding schema isn't a substitute for having clear information.
The underlying content still matters.
What About llms.txt?
There has been considerable discussion around llms.txt as websites adapt to AI systems.
Jarno takes a measured approach.
He sees llms.txt as an experiment and a helpful orientation layer, rather than a visibility button.
It may make important pages and topics easier to discover, but it cannot replace useful content, technical quality, reputation or independent evidence.
Jarno's approach is simple:
Implement sensible experiments, then measure whether they actually change anything.
That philosophy is consistent throughout his work.
Don't adopt something simply because it is currently being discussed.
Test it.
Measure it.
Then decide what the evidence tells you.
Make Pages Easy to Quote
One of Jarno's ideas is that websites should be “easy to quote, not merely easy to crawl.”
That requires thinking about how information is presented.
His practical recommendations include:
- Put the main answer where people can find it quickly.
- Use descriptive headings.
- Explain important terms clearly.
- Support claims with evidence.
- Show authorship.
- Include update dates where appropriate.
- Avoid hiding important facts inside marketing language.
The goal is to make information easier for an AI system to understand and represent accurately.
But these practices also make websites easier for humans to understand.
That overlap is important.
The AI Search Advice He Thinks Is Overhyped
The AI search industry is moving quickly, and with that comes a constant stream of new tactics.
Jarno is skeptical of the idea that one file, one prompt trick or one piece of schema can suddenly make a business “rank in ChatGPT.”
For him, AI visibility isn't a switch.
It is built from several things working together:
Helpful content + technical accessibility + entity clarity + credible references + ongoing measurement.
This is also why he prefers experimentation over blindly following trends.
If something has a plausible mechanism, can affect how people discover businesses and can be tested responsibly, he sees value in running a small experiment.
If it creates excitement but cannot be measured or explained, he treats it as noise until stronger evidence appears.
What AI Search Reveals About Businesses
One of Jarno's biggest surprises has been how uneven AI search results can be.
A business may be described accurately on one platform and poorly — or not at all — on another.
Even more interestingly, AI systems can sometimes produce confident-sounding answers when the underlying business information is incomplete or mixed up.
That creates a different kind of challenge for businesses.
It's not simply about gaining visibility.
It's about making sure the information that becomes visible is correct, consistent and supported by evidence.
Building AEOvara
Jarno founded AEOvara to help businesses navigate search in the AI era.
The challenge is partly technical and partly educational.
Many business owners already understand concepts such as Google rankings and website traffic.
But asking:
“How do AI systems describe my business?”
is still a relatively new question.
Jarno believes the answer is to keep the conversation practical.
Instead of selling mystery, show businesses:
- What AI systems currently say
- Where information is missing
- Where information is inaccurate
- What can be improved
- What changed after the work
That makes a new category easier to understand.
It also keeps the focus on measurable outcomes rather than hype.
From AI Visibility to Repeatable Benchmarks
A major part of Jarno's current exploration is making AI visibility more reproducible.
He has developed an ongoing AI visibility benchmark and published practical guides around measuring how AI systems discover and cite content.
The idea is straightforward: if businesses want to improve their AI visibility, they need a way to observe what is happening before and after making changes.
That means stable prompts, repeated observations and clearly separated metrics.
Instead of asking whether a website is “good at AI search,” businesses can begin asking more specific questions:
What changed?
Where did it change?
Was the information accurate?
Was the business cited?
Did the change lead to a meaningful outcome?
Those questions turn an abstract concept into something that can be investigated.
Lessons From Photography
Jarno's work isn't limited to SEO and AI search.
He also has a background in professional photography and visual content through Kuvaajankulma.
That experience has influenced how he thinks about discoverability and trust.
Photography taught him that presentation matters.
A website needs a clear point of view, coherent presentation and real evidence.
It also taught him to respect the user's attention.
The best page isn't necessarily the one containing the most information.
It's the one that makes the right information easy to understand.
That principle carries naturally into his thinking about AI search.
Where Search Goes Next
Jarno doesn't see traditional search disappearing completely.
He expects it to remain important, particularly for research, navigation and comparison.
At the same time, he believes AI assistants will increasingly become the starting point for many questions, especially broad or conversational ones.
Rather than imagining a complete replacement, he sees a more blended search journey emerging.
People may move between traditional search engines, AI assistants, websites and other interfaces depending on what they are trying to accomplish.
For businesses, that means the underlying information needs to work across more than one type of discovery experience.
Advice for Developers and Small Businesses
For developers, technical founders and small businesses thinking about AI search, Jarno recommends starting with fundamentals rather than chasing every new trend.
His practical starting points are:
- Keep business facts accurate and consistent.
- Make important pages technically accessible.
- Answer real customer questions clearly.
- Use relevant structured data.
- Make it clear who is responsible for the content.
- Build a stable set of prompts and measure them over time.
- Focus on changes that can actually be verified.
The common thread is simple:
Improve what you can measure.
A Lesson From Building a New Category
One lesson Jarno learned while building AEOvara is that too much jargon can make a new field harder to understand.
Technical terminology may be useful internally, but customers ultimately need answers to practical questions.
Are we being understood correctly?
Are we being recommended for the right reasons?
What changed after the work?
Those questions provide a much clearer way to explain the value of AI search visibility.
Jarno believes one of the biggest misunderstandings around AI search is treating it simply as a temporary traffic-generation trick.
He sees something broader.
AI search is also an information-quality challenge.
If a business cannot explain itself clearly, consistently and with evidence, AI systems will have a harder time understanding it correctly.
That makes AI visibility partly a technical problem, but also a content, information architecture and trust problem.
What's Next for AEOvara?
Jarno is most excited about making AI visibility more reproducible.
That includes practical benchmark methods, clearer reporting and tools that help smaller businesses understand how answer engines see them.
AEOvara has already explored this through practical guides, an llms.txt generator and ongoing AI visibility benchmarking.
The broader goal is to make AI search visibility less mysterious and more measurable.
Rather than asking businesses to trust a new acronym or a new optimization trick, the approach is to establish a baseline, run experiments, observe changes and learn from the evidence.
That mindset — measure what changes rather than simply following the latest trend — runs through Jarno's approach to AI search.
About Jarno S
Jarno S is the founder of AEOvara, based in Lappeenranta, Finland.
His work focuses on Answer Engine Optimization (AEO), Generative Engine Optimization (GEO), AI Search Optimization, Technical SEO, Schema.org structured data and LLM Optimization (LLMO).
He works across search ecosystems including Google Search, ChatGPT, Gemini, Perplexity and Copilot, exploring how AI answer engines discover, interpret and cite web content.
His current work includes entity clarity, structured data, repeatable AI visibility benchmarks and practical ways to connect traditional SEO with AEO, GEO and LLMO.
His guiding principle is simple:
“Build for people. Structure for understanding. Measure what changes.”
CoderLegion Developer Stories
This story is part of CoderLegion Developer Stories, where developers, engineers, technical founders and technology professionals share their experiences, challenges, projects and lessons from their journeys.
The goal is to give the CoderLegion community a closer look at the people and ideas behind the technology.
Interviewed and edited by Mehadi Hasan, Community & Editorial Team at CoderLegion.
