Keenable raises $26 million for AI agent web search
Keenable has emerged from stealth with a $26 million seed round led by Accel to build web-search infrastructure for AI agents. Conviction Partners and business angels also participated in the financing. The company says it has created an index of more than 100 billion documents and that its API is already used in production by several AI labs and inference providers during both model training and runtime.
The startup was founded by Andrey Styskin, formerly head of search, AI and cloud at Yandex, and German AI scientist Matthias Petri. Styskin also drew on roughly 20 years of search work at Yandex and Amazon, where he worked with Petri on infrastructure for AI applications including Alexa.
Search built around agent workloads
Keenable is pursuing the view that web search built for people does not necessarily fit AI systems. Human-oriented search has historically been designed for limited attention and quick scanning of pages. AI chatbots and agents can process much larger bodies of information, but Styskin argues that their outputs improve when answers are grounded in source documents.
Its central technical challenge is the cost of searching at internet scale. Styskin said general-purpose enterprise search approaches can break down when indexes are not tuned for a particular task: serving and scanning the full web becomes expensive because of its volume. Keenable says its work focuses on rapidly narrowing the search space for a given query, alongside proprietary retrieval capabilities.
The company recently partnered with voice AI company Gradium to support live information retrieval. It is also developing WebQueryLanguage, an upcoming product intended to help AI systems answer questions by combining information from multiple web sources when no individual source contains the complete answer.
Investment follows tighter access to search APIs
Accel partner Zhenya Loginov, who led the investment, said AI companies have few choices for web-scale search infrastructure. Google and Microsoft have taken steps to shut down existing search APIs in order to avoid cannibalisation, while moving toward more bundled offerings and selecting partners more selectively.
Styskin said those decisions reinforced an opportunity he had identified while working on AI search infrastructure at Amazon. He had also seen Cloudflare data indicating that AI crawlers accounted for a growing share of search volume. Keenable now joins other companies targeting the space, including Brave and Exa, while Google continues to reshape its own search experience for the AI era.
Scaling the index and commercial operation
Building a web index at this scale remains costly. Styskin described the expense as painful, while saying the company is seeking to keep costs under control and expand at a measured pace. Keenable has 15 engineering staff across the United States and Europe and plans to double its headcount by year-end, using the new funding to establish its go-to-market motion.
For businesses building agentic products, the practical question is not simply whether an agent can search the web, but whether retrieval can be grounded, fast and affordable at production scale. Teams evaluating such systems should examine their source coverage, retrieval design and runtime cost before making web access a core product dependency.