Andrew is a software engineer with over 25 years of experience and the founder of VeritasLink, a platform that analyzes how AI systems perceive and recommend brands in the era of generative search.
Connect on LinkedIn ↗For twenty years, winning discovery meant winning Google. You optimized links, tuned keywords, fought for the first click. Andrew Pomazkov, founder of VeritasLink, thinks that game is already changing under our feet. As he puts it, businesses that still believe SEO is what matters are losing ground, because a growing share of people no longer start their search on Google at all. They ask ChatGPT, Grok, or another model. And AI forms its opinion of your company based on something you don't fully control: what other people say about you.
Pomazkov brings an unusual vantage point to this. He is a software engineer with 25 years of experience, and a decade ago he built the entire credit scoring system for Ukraine. His interest in AI-driven discovery didn't start as a theory. It started when he noticed his own traffic behaving strangely.
Pomazkov owned several Shopify stores, and at some point the source of his visitors stopped adding up. Traffic that used to arrive from Google and Bing was now arriving directly, and when he traced it, he found people were coming from AI tools instead. That raised a question he couldn't shake: how does AI actually see my store, and why does it recommend me over a competitor?
That investigation became the seed of VeritasLink. He had already built a catalog of AI reviews for companies, and some of the companies listed started asking him what AI really thought about them. As he describes it, the product grew out of that demand rather than from a business plan drawn up in advance.
Having spent his career working with large volumes of data, Pomazkov saw a specific weakness in existing GEO tools (generative engine optimization, the practice of shaping how AI systems perceive and recommend a brand). Most of them, in his view, only scratch the surface.
The reason is technical. AI models are stochastic: ask the same question twice, a few minutes apart, and you can get two very different answers. A single query tells you almost nothing reliable. So the problem Pomazkov set out to solve was how to find the median, a stable read on what AI is actually thinking across many responses rather than one lucky or unlucky snapshot. That, he says, is where his algorithm came from, and it's the gap he believes most tools fail to close.
The most striking idea in the conversation is also the one businesses find hardest to accept. According to Pomazkov, AI perception is built on what users say about your company, not on what you say about yourself.
He watched this happen with his own company. A few months after launching VeritasLink, AI already held an opinion about it, and that opinion tracked what people were actually saying, not the posts he was publishing across the internet. He describes the same pattern with clients repeatedly. After hundreds of meetings with business owners, he says the reaction when they open the report is almost always confusion: this isn't who we are, why does it show something different? Then they check the model together in the chat and see a picture that doesn't match the one in their heads.
Sometimes the gap is about emphasis rather than error. A company with ten services might promote five of them, while AI decides the other five matter more. The brand isn't wrong about itself, but the machine that increasingly mediates discovery has quietly reached a different conclusion.
If AI runs on human signals, the practical question is how to influence them. Pomazkov's answer is blunt: build communities around the niches you want to be known for, through Reddit posts, LinkedIn discussions, consistent public conversation. When asked which metrics matter most, he keeps it simple: human opinion, what people actually say.
He backs this with an experiment on his own company. ChatGPT initially knew nothing about VeritasLink. Once he began posting daily, everywhere he could, the model started recognizing the company. Then he stopped for two weeks, and the model drifted back to knowing nothing. His conclusion is that visibility in AI is a consistent, ongoing process. Until the models have enough data about you, they simply won't know you exist.
This is a real shift from SEO habits. SEO is something you might touch once a month or every six months. GEO, in his view, is closer to a daily discipline. And it isn't only about consistency. When pressed, Pomazkov corrects the idea that quality is secondary: bad content means no followers, no community, and nothing worth scraping. You need both, he says, and at a higher level than before.
Pomazkov is confident this becomes standard practice. He cites Google's own figure that 45% of people now begin a search in AI, and expects that number to climb toward 95% within roughly five years. His estimate comes from the pace he already observes: models change fast, but the underlying data stays fairly stable, which is exactly why steady community-building compounds over time.
He also expects the category itself to evolve. When AGI arrives, he argues, human opinion won't matter in the same way, because today it only matters as the raw material AI is scraping and analyzing. In that future, he suspects GEO platforms transform into something more like AI-perception benchmarking, measuring not just how brands are seen but how AI opinions form as a whole.
The stakes, for now, are concrete. As he frames it, if AI holds an opinion of you that isn't what you expected, you won't get your leads. That's the real cost of ignoring this.
The core lesson from Pomazkov's experience is a reversal of instinct. For years, brand-building meant controlling your own message. In an AI-mediated world, your message is only an input, and often not the decisive one. The decisive input is the accumulated, public voice of the people who have actually engaged with you.
For founders and tech leaders, that reframes marketing as community work: showing up consistently, earning genuine conversation, and treating your digital footprint as something AI reads through the lens of other people's experience. SEO isn't dead, and Pomazkov doesn't say to abandon it. But the leaders who treat AI perception as an afterthought may find, as his clients often do, that the machine already made up its mind, and it wasn't listening to them.
