Nineteen years is a long time in search. Long enough for Google to rewrite its own rulebook several times, for smartphones to become the default screen, and for a new kind of search engine to arrive that answers questions without sending anyone to a website at all. CRS AI Marketing & SEO Agency has lived through all of it. Its story is a useful map of how digital visibility changed, and of why its founder ended up building an AI consultancy and a research lab on top of it.

When and why did Miklós Róth found the agency?

Róth founded the company in 2007, after a decade that had taken him from international athletics through finance studies in Nebraska, business ventures in Spain and the United States, and team leadership at IBM. Registered today as CRS AI Marketing & SEO Ügynökség Kft. (formerly CRS Budapest Kft.), it is based at Jászai Mari tér 5–6 in Budapest's 13th district.

The original bet was simple. Hungarian companies were starting to compete online, and most of them had no idea how search engines decided who got seen. In 2007 the answer was mostly about rankings: ten blue links on a results page, and if you weren't in the top three, you barely existed.

How has search changed since 2007?

Every step on that staircase changed what an agency had to deliver.

2007–2012: rankings and links. Visibility was a game of keywords, on-page optimization and backlinks. It was measurable, sometimes crude, and it rewarded technical persistence.

2013–2017: meaning over keywords. Google's Hummingbird update in 2013 pushed search toward understanding intent and topics instead of matching strings. Content quality, site structure and entities started to matter more than keyword density.

2018–2021: the phone is the web. With mobile-first indexing, rolled out from March 2018, Google began treating the mobile version of a site as the primary one. Speed, technical health and local visibility became core agency work, not extras.

2022 onward: answers without clicks. ChatGPT's launch in late 2022, followed by Google's AI Overviews in 2024, changed the unit of success. A brand can now be recommended, or ignored, inside an AI-generated answer without the user ever seeing a results page.

What does CRS do today?

Today the agency's work runs along three main lines.

AI strategy and business transformation. Helping companies integrate AI-based systems and tighten their decision processes. This is the bridge to Róth's advisory work at Roth AI Consulting.

Classic and AI-extended search optimization. Technical SEO, semantic SEO and E-E-A-T, plus Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO): structuring a brand's content and reputation so that AI systems understand it, trust it and cite it. We walk through the method in How Miklós Róth Gets Brands Cited by AI.

Predictive and AI-assisted marketing. Using AI not just to produce content and traffic but to support forecasts on revenue, retention and churn, alongside AI-assisted premium link building and content production at scale.

Clients include Hungarian small and mid-sized businesses, B2B manufacturers and service firms, and webshops, together with clients elsewhere in Europe and in the United States.

What results has the agency published?

The agency's case studies and partner press coverage cite concrete outcomes. For industrial technology company Modern Ipartechnika Kft., the agency reports a 450 percent rise in qualified inquiries within eight months, plus a project worth 120 million forints that arrived through organic search. For the Szeptest brand, a TechBullion feature describes a 200 percent increase in visibility and a 40 percent cut in customer acquisition cost after AI-driven keyword forecasting and audience segmentation.

The pattern matters more than any single number. The target is never traffic for its own sake. It is inquiries, revenue and cheaper acquisition, measured in the client's own pipeline.

What hasn't changed since 2007?

Quite a lot, which is reassuring for anyone overwhelmed by AI search. Search engines and language models still reward the same three things they always have, only more strictly.

Being technically reachable. A page that can't be crawled, loads slowly or hides its facts inside images was invisible in 2007, and it is invisible to AI systems now.

Being genuinely useful. Content that answers a real question clearly beats content written to fill a keyword slot. AI systems are simply better at telling the difference.

Being trusted by others. Links, mentions, reviews and press coverage were ranking signals then. Today they also tell a language model whether your claims are corroborated.

The tactics changed every few years. The fundamentals never did.

Why does an agency founder run a research lab?

Because client work kept raising questions that marketing playbooks couldn't answer. Why do some companies absorb a new channel or tool smoothly while others descend into chaos doing exactly the same thing? Why does publishing more content sometimes make a brand less visible, not more?

Róth's answer became the S·I·C·T framework, developed at the Roth Complexity Lab: visibility, like any property of a system, breaks down when information and change outrun structure and cohesion. A site that publishes a hundred AI-generated articles without a clear entity, structure or internal logic is adding information without adding structure. Search engines and language models notice. The framework is explained here.