Keenable Raises $26M for AI Agent Search Infrastructure

The next search customer may never see a results page. Keenable has raised a $26M Seed round led by Accel, with Conviction and business angels participating, to build web-search infrastructure for AI labs and software agents.

The San Francisco company is betting that a machine making hundreds or thousands of retrieval calls needs a different index, cost model, and ranking system from a person scanning 10 blue links. The financing gives Keenable more capacity to grow its engineering team and commercial operation, but it also creates a demanding proof obligation: the startup must keep an independent index fresh, fast, useful, and economically sustainable while competing with both dedicated retrieval companies and the platforms that already organize the web.

What Happened

Keenable emerged from stealth on August 25, 2026 with a $26M Seed financing led by Accel. Conviction and unnamed business angels also participated, while valuation and detailed ownership terms were not disclosed. TechCrunch reported that Accel partner Zhenya Loginov led the investment.

Keenable was founded in 2025 by Andrey Styskin, co-founder and CEO, and Matthias Petri, co-founder and Chief Scientist. The official company site says Styskin previously led Search and Ads at Yandex and later served as Director of Web Infrastructure at Amazon AGI. Petri was a Principal Applied Scientist at Amazon AGI, and Keenable says he helped build the web-grounding service behind Alexa.

Why AI Agents Need Different Search Infrastructure

Consumer search was optimized around limited human attention. A person enters a query, considers a small results set, opens a few pages, and decides whether the answer is useful. An AI agent can make retrieval one step inside a longer workflow, inspect far more source material, combine evidence, and query the web again before acting.

That workload changes the commercial problem underneath search. Latency becomes part of the agent's response time, retrieval cost compounds across repeated calls, and source quality matters because the final answer may be delivered without a human inspecting the original page. Keenable is selling into that architectural pressure rather than trying to win a consumer homepage.

What Keenable Has Built

Keenable says its independent index covers more than 100B documents and that its API is already used in production by several AI labs and inference providers during training and runtime. Those customers remain unnamed, and the coverage, performance, and adoption figures are company-reported rather than independently audited. Keenable's website also advertises p95 latency below 250ms in US East and pricing from $1 per 1,000 requests at high throughput.

The company currently promotes a Search API for agents and a Time Machine product that searches historical versions of web pages. A July partnership with Gradium offers one public example: a voice agent retrieves fresh web information while maintaining conversational response time. Keenable says its search returns in under 200ms in that integration, an operating claim that still requires customer-level validation at broader scale.

Why the Market Opening Exists Now

The established search platforms are narrowing standalone access while expanding their own AI products. Google's Custom Search JSON API is closed to new customers, and existing customers must migrate before the service is discontinued on January 1, 2027. Microsoft retired the legacy Bing Search APIs on August 11, 2025 and directs customers toward Grounding with Bing Search inside Azure AI Agents.

That does not leave Keenable alone. Brave and Exa already sell search and retrieval for AI applications, while other infrastructure companies are competing across crawling, ranking, answer synthesis, and agent tooling. The opening is real, but an open market is different from an open lane.

What the $26M Must Buy

TechCrunch reports that Keenable has 15 engineers across the United States and Europe and plans to double headcount by the end of 2026 while building its go-to-market motion. The capital therefore supports both technical scale and customer acquisition, two expenses that can become inseparable when buyers need proof before they move a production agent onto a new index.

Infrastructure economics will shape the company's room to maneuver. Crawling, storing, refreshing, ranking, and serving a meaningful portion of the web is capital intensive, and Keenable must offer developers predictable quality and cost without starving the index that creates the product's value. The round funds the attempt, but the customer experience will decide whether the index becomes an asset or an expensive claim.

The Proof Keenable Still Owes

Keenable's strongest early evidence is the founders' search background, a named Gradium integration, investor conviction from Accel and Conviction, and the company's assertion that multiple AI labs already use the API in production. Its biggest gaps are equally clear: customers are unnamed, the 100B-document coverage is self-reported, and public evidence does not yet establish how freshness, precision, recall, duplication, source quality, and unit economics behave at sustained production volume.

AI builders should read the round as a signal that the web index beneath an agent is becoming a commercial and architectural choice. Keenable now has $26M to make independent retrieval credible at machine scale. The next proof will arrive when customers can show that the index stays fresh, the costs stay rational, and the answers deserve the confidence placed on top of them.