Keenable raises $26M for a web index built only for agents

Startup Keenable came out of stealth on August 25, 2026, announcing a $26 million seed round led by Accel with participation from Conviction Partners. The product: a proprietary web search index of more than 100 billion documents — built not for human users but as an API for AI agents, AI labs and inference providers.

Why does this need a new index? Classic search engines rank for people: ten blue links optimized for clicks. Agents instead issue long, nested queries and combine information from many sources. For that, Keenable is developing its own query language, WebQueryLanguage, which lets AI systems compose answers across multiple web sources.

Co-founder Andrey Styskin — previously a senior search executive — sees a structural opening: "This actually creates a new flywheel that is different from what Google learned from human behavior." Asked about the cost of building the index, his answer was dry: "Don't ask — it is painfully expensive."

How far along is the company? Several unnamed AI labs already use the index in production, Keenable says. The team currently counts 15 engineers and is set to double by year-end. That puts Keenable in a growing group of infrastructure startups rebuilding search for the agent era — Seltz, for instance, recently raised $12.5 million for a similar goal.

The trend matters for European companies too: if agents increasingly do their own research, the underlying search infrastructure helps decide which sources shape AI answers — and who controls access to them.