Keenable's $26M Bet on Search Engines for AI

Key points

• Keenable raised $26 million in seed funding, led by Accel.
• Its web index covers more than 100 billion documents.
• The API is already running in production at AI labs and inference providers.
• The startup plans to double its 15-person engineering team by year end.

Why build a search engine that ignores humans?

Search engines were designed for people who skim. AI agents don't skim: they consume. Andrey Styskin, who led search and AI at Yandex for two decades before working at Amazon, co-founded Keenable with German AI scientist Matthias Petri on the conviction that the web's infrastructure needs rebuilding with AI agents as the primary reader. The $26 million seed round was led by Accel, with Conviction Partners and several business angels also participating.

"This actually creates a new flywheel that is different from what Google learned from human behavior," Styskin told TechCrunch.

How does Keenable's index work?

The company has built a web index of more than 100 billion documents. Its API is already live in production at AI labs and inference providers, covering both training and real-time use. Keenable won't name its customers, but it has publicly confirmed a partnership with voice AI company Gradium for live information retrieval.

Styskin's pitch on cost is blunt. Without fine-tuning index structures for specific tasks, "the cost of serving and scanning the whole internet is enormous because of the volume," he told TechCrunch. Keenable's answer is faster narrowing of the search space per query. An upcoming product called WebQueryLanguage would let AI systems combine partial answers from multiple sources when no single page holds the full picture.

For what it's worth, Accel partner Zhenya Loginov noted that Google and Microsoft have both been pulling back their search APIs to protect their own bundled AI products, which hands a real opening to independents. We covered Accel's recent moves in its $550 million India fund raise on 11 August.

What are the challenges ahead?

Building a web-scale index costs a lot. "Don't ask, it is painfully expensive," Styskin told TechCrunch. Keenable's 15 engineers are spread across the US and Europe, and the new cash is earmarked to double that headcount by year end.

Competitors include Brave, Exa, and Google itself, which is overhauling its own search experience for AI. The broader point is that Google's dominance on agentic queries is less entrenched than on human search, because Google hasn't optimised for agents the way it has for people. That's where Keenable is placing its bet.

Common questions

Why is AI search different from regular search?

AI agents can process far more text per request than a person reads per session, so they need an index built for high-volume, structured retrieval rather than ranked links designed for a human skimming ten results.

What does this mean for traditional search engines?

Engines built purely for human users face pressure from two directions: AI products that bypass them entirely, and customers who expect cheaper, more direct data access than bundled platforms offer.