Keenable, a startup building web search infrastructure for AI agents, has emerged from stealth with a $26 million seed round led by Accel, according to TechCrunch. Conviction Partners and several angel investors also participated in the round. TechCrunch reports that the company was founded by Andrey Styskin, who previously led Yandex’s search, artificial intelligence and cloud division, and Matthias Petri, a German AI scientist. The company’s premise is that search infrastructure built for human users is not necessarily the right foundation for AI systems that can ingest and reason over far more source material than a person would scan in a results page. Keenable says it has built a web search index of more than 100 billion documents. According to TechCrunch, the company says its application programming interface is already being used in production by several AI labs and inference providers, both during training and at runtime. Keenable did not disclose those customers. The startup has named at least one partner: TechCrunch reports that Keenable recently struck a partnership with voice AI company Gradium to support live information retrieval. That detail matters because agentic and voice AI products often need current external information rather than only model-embedded knowledge. Styskin told TechCrunch that AI chatbots perform better when their responses can be grounded in source documents. He argued that this creates a different feedback loop from the one search engines learned from human behavior. The company is positioning its index and retrieval layer around that difference: fewer assumptions about what a person can read, and more emphasis on rapidly narrowing a very large web-scale search space for machine consumption. TechCrunch also reports that Keenable is building proprietary retrieval capabilities, including an upcoming product called Web Query Language. The described aim is to help AI systems answer questions by combining information from multiple web sources, including cases where no single source contains the full answer. The competitive opening, as described by Accel partner Zhenya Loginov to TechCrunch, is that AI companies have limited choices for web-scale search infrastructure. Loginov said Google and Microsoft have been taking steps to shut down existing search application programming interfaces to avoid cannibalization, while moving toward more bundled and selective partner approaches. That claim should be treated as investor framing, but it explains why Accel sees room for an independent search layer aimed at AI developers. For now, the strongest confirmed facts are the financing, the founders, the index claim, and Keenable’s stated product direction. The important unanswered questions are commercial: who the unnamed AI lab and inference-provider customers are, what performance or cost advantages Keenable can prove at scale, and whether developers will move critical retrieval workloads away from incumbent search providers. Who benefits: Keenable benefits from a large seed round and Accel’s backing as it tries to sell search infrastructure to AI companies. AI labs, inference providers and voice AI developers could benefit if the API gives them a workable way to ground model outputs in web sources. Who's exposed: The reporting points to pressure on incumbent web-scale search/API providers such as Google and Microsoft. Keenable’s own exposure is execution risk: web-scale indexing and retrieval are expensive, and its customer traction is still mostly unnamed.