Skip to content
NewsFunding & M&A

Keenable raises $26M from Accel to build a web search index designed for AI agents

· by Pondero Newsdesk

The short version

Keenable exited stealth on August 25 with a $26 million Accel-led seed round and a 100-billion-document index built for the high-volume retrieval patterns of AI agents, not human queries.

Keenable raises $26M from Accel to build a web search index designed for AI agents

As Google and Microsoft pull back on open search APIs, Keenable emerged from stealth on August 25 with $26 million in seed funding and a proprietary 100-billion-document index built specifically for how AI agents consume the web.

What happened

Keenable was co-founded by Andrey Styskin, who led Yandex's search, AI, and cloud division for nearly 20 years and later worked on web search infrastructure at Amazon, and Matthias Petri, a German AI researcher. Accel partner Zhenya Loginov led the $26 million seed round, with Conviction Partners and several business angels participating, per TechCrunch.

The company's core product is a web search API backed by its own index of more than 100 billion documents. The API was already running in production at several unnamed AI labs and inference providers when Keenable came out of stealth. Keenable also struck a partnership with voice AI company Gradium to support live information retrieval.

Styskin's argument, explained to TechCrunch, is that existing search indexes were engineered for single human queries: short dwell time, one result page, no iteration. AI agents run differently. A multi-step research task might fire hundreds of retrieval calls in sequence, each building on the last. Keenable's index structure and cost model are optimized for that pattern. An upcoming product, Web Query Language, would take the platform a step further: letting AI systems synthesize an answer by combining partial information from multiple web sources, rather than surfacing individual pages.

The company employs 15 engineers across the US and Europe. The seed capital will fund index expansion and doubling the engineering headcount by the end of 2026, per TechCrunch.

Why it matters

The funding thesis rests on a real infrastructure gap. Google recently began shutting down its Custom Search API, and Microsoft has announced the retirement of the Bing Search API. Both decisions reflect the incumbents' preference for bundled cloud deals over open, pay-per-call access, according to Loginov's comments in the TechCrunch report. That leaves operators building retrieval-heavy agents with a narrow set of independent options: Brave Search API, Exa, and now Keenable.

Keenable's pitch is that cost-per-call economics at high agentic volume are different enough from human-search economics to justify a purpose-built index. Styskin told TechCrunch that Google remains hard to displace for human search, but argued that agentic queries are where an agile, cost-focused competitor can undercut and out-innovate. Whether that holds at scale depends on latency and freshness benchmarks that the company has not published.

For AI builders selecting a web grounding provider today, Keenable's entrance gives the market a third serious independent option. The fact that its index is already in production at unnamed AI labs suggests real-world validation, not just a launch-day demo.

What to watch next

Keenable's paying customers remain under NDA. Naming even one would sharpen the market's read on where the index sits in the performance stack relative to Brave and Exa. The Web Query Language product is the next milestone: if it ships, Keenable moves from search infrastructure to a higher-level reasoning layer, and the competitive comparison set shifts again.

Sources