Keenable, a Los Angeles-based search infrastructure startup, announced Tuesday it raised $26 million in seed funding led by Accel. The round included Conviction Partners, Brightwing Capital, ScOp Venture Capital and executives from Google and Amazon, according to a report in TechCrunch.
The company, which incorporated just months ago, emerged from stealth claiming to have indexed more than 100 billion documents. It declined to disclose its valuation.
The timing is no accident. Google closed its Custom Search JSON API to new customers in February, giving existing users until early next year to migrate, according to developer documentation. Microsoft retired the Bing Search APIs in August 2025, redirecting developers toward Azure AI Agents instead. Those shutdowns left AI labs scrambling for production-grade web access. Keenable thinks it can fill that gap.
At its core, the startup maintains what it describes as a large, independent web index. Through a search API, it delivers what it calls "model-ready" inputs—structured data designed to drop directly into AI systems. The API supports command-line interfaces, REST endpoints and Model Context Protocol server integrations. There's also a "Time Machine" feature that lets developers query historical snapshots of the web, though access to that remains limited.
Keenable said several AI labs and inference providers already rely on the API in production but wouldn't name them. In a blog post from July, the company claimed it can deliver web retrieval in under 200 milliseconds, fast enough to sit inside an agent's reasoning loop. It also announced a partnership with Gradium, a voice AI startup, to power live retrieval during natural conversation.
The company was founded by Andrey Styskin, who previously ran Search & Ads at Yandex, the Russian search engine. In a biography on the company's site, Styskin described that business as generating $2 billion in revenue and outperforming Google Search in Russia. Before launching Keenable, he served as Director of Web Infrastructure at Amazon AGI, according to his LinkedIn profile.
Co-founder Matthias Petri came from Amazon as well, where he worked as an applied scientist on Alexa and AGI teams. His research on retrieval and indexing has appeared in SIGIR and EMNLP conferences, according to his Amazon Science author page.
When TechCrunch asked Styskin how much it costs to build a 100-billion-document index, his answer was succinct: "Don't ask—it is painfully expensive."
Pricing runs $4 per 1,000 requests on the standard tier, dropping to $1 per 1,000 for customers hitting 100 requests per second or higher, according to the company's pricing page. Developers get 100,000 free requests per month to start. That sits below Perplexity's $5 per 1,000 searches and roughly in line with Brave Search API, which charges around $5 per 1,000 searches depending on the plan. Exa, another API-first provider, lists pricing that varies widely, often in the sub-cent-per-request range. Tavily markets agent-focused search and content extraction using a credit-based model.
Keenable now employs 15 engineers spread across the United States and Europe, with plans to double headcount by year-end, according to a report in SiliconANGLE. The company's careers page lists openings for a Head of Product, Head of Revenue, GTM/Scientist Relations and a senior engineering role, all remote.
SiliconANGLE also reported that Keenable is developing proprietary retrieval models and an upcoming "Web Query Language" meant to combine data across sources, though the company hasn't publicly detailed those plans.
"This is what we are bringing to the table," Styskin told TechCrunch, describing Keenable's ability to quickly narrow the search space for AI agents working through complex, multi-step tasks.
Accel partner Zhenya Loginov led the deal. "Today, we announced we're leading Keenable's seed round as the company emerges from stealth," Accel wrote in a LinkedIn post the following day.
