Keenable Raises $26M for Agentic Web Search

TechCrunch and SiliconANGLE report that Keenable's index covers more than 100 billion documents. SiliconANGLE describes APIs for natural-language search and cleaned-content fetching, targeting workloads including market mapping, price monitoring, and lead enrichment. The outlet reports that the service is priced at $1 per 1,000 API requests for frontier-scale users.

A notable feature is point-in-time retrieval. According to SiliconANGLE, Keenable supports historical queries against the web as it existed at a specified time, rather than only its current state. For model developers, this capability can be relevant to reproducible evaluation, training-data provenance work, and retrieval tasks where the time of a fact matters.

Styskin told TechCrunch that AI chatbots can perform better when their answers are grounded in source documents. In comments reported by SiliconANGLE, he described task-specific index structures as necessary to reduce the cost of scanning web-scale content: "If you do not fine-tune your index structures for a specific task, the cost of serving and scanning the whole internet is enormous because of the volume."

TechCrunch reports that Keenable's API is already in production with several AI labs and inference providers during both training and runtime, though the startup did not disclose customer names. FinSMEs separately reports that Keenable has secured commercial contracts with multiple AI labs and is working with model builders and inference providers to connect AI systems to live web data.

FinSMEs reports that the startup intends to use the funding to expand its engineering team across the Bay Area and Europe, grow its web-indexing and crawling infrastructure, and continue developing the retrieval systems underlying its Web Query Language.

The product category sits at a consequential layer of the agent stack: retrieval quality, latency, freshness, and unit cost can constrain tool-using systems even when the underlying language model is strong. Companies building comparable web-scale retrieval systems typically need to balance broad crawl coverage against query-time filtering, document cleaning, ranking, and citation-quality source selection. Keenable's reported emphasis on task-specific indexing addresses that general systems problem, while independent measurement of recall, freshness, latency, and cost across workloads remains an open question for prospective users.