Keenable Raises $26 Million for 100 Billion Document AI Search Index
Accel-backed Keenable raises $26 million to build a 100 billion-document web index and cost-efficient retrieval infrastructure for AI agents at web scale.
Summary
On August 25, 2026, Keenable emerged from stealth with a $26 million seed round led by Accel partner Zhenya Loginov, with Conviction Partners and business angels participating. Founders Andrey Styskin, formerly head of Yandex’s search, AI and cloud division, and German AI scientist Matthias Petri previously built web search infrastructure for AI applications including Alexa at Amazon. Styskin brings 20 years of search experience across Yandex and Amazon. Keenable’s 15 engineers across the U.S. and Europe will double by year-end to build its go-to-market operation.
Keenable is building an index exceeding 100 billion web documents. Its API already supports undisclosed AI labs and inference providers during training and runtime, while voice AI company Gradium uses it for live information retrieval. Proprietary retrieval technology narrows each query’s search space to reduce web-scale scanning costs. The upcoming Web Query Language will combine multiple sources when no single page contains a complete answer, helping AI systems ground responses in source documents.
Google and Microsoft are moving to close existing search APIs, avoid cannibalization and favor bundled access with selected partners, while Cloudflare data shows AI crawlers taking a growing share of search volume. Styskin believes Google may be beatable on agentic queries through cheaper infrastructure, although maintaining a vast index remains painfully expensive. Keenable aims to become the next Google for AI agents but faces Brave, Exa and Google’s own AI search overhaul.
Positives
- $26 million in seed funding gives Keenable capital to expand its search infrastructure and commercial operation.
- More than 100 billion indexed documents provide web-scale coverage for AI training and runtime queries.
- Several AI labs and inference providers already use Keenable’s API in production, despite their identities remaining undisclosed.
- Gradium’s partnership validates Keenable’s live retrieval capabilities for voice AI applications.
- Web Query Language will let AI systems synthesize answers from sources that each contain only partial information.
- Keenable plans to double its 15-person engineering team across the U.S. and Europe by year-end.
Risks & concerns
- Maintaining an index exceeding 100 billion documents is painfully expensive, making cost control central to Keenable’s prospects.
- Google, Brave and Exa already compete in AI-focused search, while Google is overhauling its own search experience.
- Google and Microsoft are restricting existing search APIs and favoring bundled access with selected partners.
- Keenable has not identified the AI labs and inference providers using its API in production.
- Web Query Language remains an upcoming product, with no launch date or performance details disclosed.