Key Takeaways：Search startup Keenable is coming out of stealth and has raised a $26 million seed round. Key details:
- Reported date：On August 25, 2026, TechCrunch reported that it has exited stealth mode
- Round：Seed, $26 million
- Lead investor：Accel
- Participants：Conviction Partners and a number of angel investors
- Product status：Its API is already in production use at several AI labs and inference providers, covering both training and runtime scenarios
- Recent partnership：Recently partnered with voice AI company Gradium to support real-time information retrieval
- Team size：Currently has 15 engineers, plans to double headcount this year with the new capital
Technical Approach and Differentiation Keenable was founded by Andrey Styskin, formerly head of search, AI, and cloud at Russian search giant Yandex, and German AI researcher Matthias Petri. The two previously worked together at Amazon on the web search infrastructure required for AI applications like Alexa—an experience that shaped much of their founding thesis.
The company is building dedicated web retrieval infrastructure for AI agents, which differs significantly from traditional search engines designed for human users. Human search relies heavily on result summaries and quick scanning, while AI agents can read and process far larger volumes of web content—meaning index structures, query routing, and retrieval methods may all need to be rethought. Styskin told TechCrunch this could create a new flywheel distinct from the one Google built around human behavior.
One standout fact is scale: Keenable claims to have built a web search index covering more than 100 billion documents. That scale exceeds the typical use cases for many traditional enterprise search solutions and underscores the infrastructure need that’s specific to AI. As Google and Microsoft have moved to restrict their existing search APIs to avoid cannibalizing their own businesses, third-party companies like Keenable are stepping in to fill the gap in web-scale retrieval infrastructure. “There are very few options for AI companies when it comes to web-scale search infrastructure,” noted Accel partner Zhenya Loginov.
Cost and Architecture Innovation Styskin stressed that serving and scanning at web scale without finely tuning index structures for specific tasks would be prohibitively expensive—the challenge comes from both the size of the index and the breadth of the search space that queries must cover. He explained that Keenable’s core capability lies in “how quickly we can narrow the search space given a query,” which requires simultaneous optimization across index structure, retrieval strategy, and query planning.
He also acknowledged that building a massive index is “painfully expensive,” but the company is working hard to control costs and manage its growth pace. Keenable is also developing a product called Web Query Language, aimed at helping AI systems synthesize information from multiple web sources to answer questions even when no single source contains the complete answer.
Market Position and Competitive Landscape Styskin believes that while “it’s incredibly hard to convince users to leave Google Search,” Google may still face challenges in agent-driven query scenarios when viewed through the lens of the “innovator’s dilemma.” For smaller companies, there’s an opportunity to carve out a niche in the infrastructure market by delivering more efficient, cost-controllable solutions tailored to agent queries.
Other players are already moving in this direction, including Brave and Exa; Google itself is reimagining search for the AI era. As TechCrunch observed at the end of its piece, the era of “ten blue links” may be coming to an end—whether the user is human or agent.
Who Should Care
- Good fit for: AI companies building agent applications that need real-time web retrieval, multi-turn reasoning systems, or long-document understanding products; AI teams that want to operate independently of the Big Tech ecosystem and customize their own retrieval strategies
- Consider waiting: Enterprise customers with very high requirements for data freshness, language coverage, compliance, or production stability may want to watch for more customer case studies and long-term operational track records
Final Thoughts Search infrastructure may be heading toward a new wave of specialization, following the precedents set by cloud computing and the API economy. As AI agents become the new entry point for information consumption, the degree to which retrieval services become industrialized and toolified will shape the efficiency ceiling of the entire agent economy. Keenable’s emergence is one of the early signals of this trend.
