Google's Search Dominance Faces a Machine-Age Challenge

A startup backed by Yandex's former search chief just raised $26 million to build search infrastructure designed for AI agents, not humans.

Google was engineered for a specific user: humans with a 30-second attention span. Ten blue links, a quick skim, a click. But the internet's primary user is shifting. Increasingly, it's AI chatbots retrieving information, processing data, and completing tasks at scales that make human search habits look quaint.

The problem is structural. Chatbots don't need snippets or ranked link lists. They can absorb hundreds of pages where a person would scan a headline. The entire infrastructure of web search — from indexing to ranking to display — was optimized around human eyeballs. One startup is now betting that this mismatch is the vulnerability in Google's seemingly unbeatable dominance. Watch the full video or listen to the podcast episode.

The Vacuum at Scale

Meet Keenable. Founded by Andrey Styskin (who spent two decades leading search and AI at Yandex, Russia's search giant) and AI scientist Matthias Petri, the company raised $26 million in seed funding led by Accel, with Conviction Partners and angel investors also backing the bet. Their thesis: the web's current search infrastructure was optimized around human behavior, but AI agents need something fundamentally different.

The timing is urgent. According to Accel partner Zhenya Loginov, who led the investment, AI companies have extremely limited options for web-scale search. Google and Microsoft are actively restricting their existing search APIs—specifically to avoid cannibalizing their own products—and being highly selective about partnerships. Dozens of AI labs and startups need to search the web constantly. The two companies that historically owned that infrastructure are pulling up the drawbridge.

What They Actually Built

This isn't a pitch deck. Keenable has already indexed over 100 billion documents and deployed its API in production with multiple AI labs and inference providers. They've confirmed at least one customer: Gradium, a voice AI company, uses Keenable for live information retrieval during conversations—real-time answers, not pre-baked responses.

The innovation isn't just scale; it's speed. Enterprise search tools crumble at web scale. As Styskin explains, without fine-tuning index structures for specific tasks, serving the entire internet becomes prohibitively expensive simply due to volume. The real product is narrowing the search space intelligently based on each query. And there's more coming: Web Query Language, an upcoming product designed to let AI systems assemble answers by pulling together information from multiple web sources—answers that might not exist on any single page.

The Incumbents' Dilemma

Styskin is candid about the challenge: convincing people to abandon Google for search is extraordinarily difficult. Google isn't just a habit; it's a reflex. But he sees a crack in the armor. Google has an existing human-search business it cannot afford to cannibalize. A smaller, focused company moving faster and cheaper in the narrow lane of agentic queries—complex, multi-step searches that AI agents run—could outmaneuver the incumbent.

That said, this isn't a low-cost venture. Styskin's response when asked about the expense of building a giant search index:

"Don't ask—it is painfully expensive."

Keenable isn't alone. Brave and Exa are moving in the same direction. Even Google itself is rebuilding its search experience for the AI era—arguably the loudest signal of all. If the company that invented the ten blue links is dismantling that paradigm, the era of human-optimized search might not be ending in a startup pitch. It might be ending in Google's own roadmap.