Accel-backed Keenable is indexing the web for AI agents

Traditional search engines were engineered for a human-centric era, designed to provide quick snippets for users who lack the time to parse through vast amounts of raw data. However, as the digital landscape shifts toward AI-driven autonomous agents capable of consuming and synthesizing massive volumes of information, the underlying infrastructure of the internet is beginning to show its age.

Enter Keenable, a new startup emerging from stealth today with a significant $26 million seed round. Led by Accel, with participation from Conviction Partners and various angel investors, the company is positioning itself to become the foundational search layer for the next generation of AI.

Rethinking Search for the Machine Age

The startup is the brainchild of Andrey Styskin, a veteran who previously spearheaded search, AI, and cloud initiatives at the Russian search giant Yandex, and Matthias Petri, a prominent German AI scientist. Their mission is to solve the fundamental inefficiency of current search tools when they are utilized by AI rather than humans.

According to Styskin, the current "flywheel" of search is built on human behavioral patterns, which is fundamentally misaligned with the needs of AI models. By grounding AI responses in high-quality, relevant source documents, Keenable aims to create a new, more efficient paradigm.

“This actually creates a new flywheel that is different from what Google learned from human behavior,” Styskin noted.

A Massive Index for AI Workflows

Keenable has already constructed a proprietary web search index comprising over 100 billion documents. This infrastructure is currently being utilized in production environments by several AI labs and inference providers, supporting both training phases and live runtime queries. While the company is keeping its client list confidential, it recently announced a strategic partnership with voice AI firm Gradum to facilitate real-time information retrieval.

Styskin, who spent two decades refining search technology at Yandex and Amazon, emphasizes that Keenable’s architecture is distinct from standard enterprise search solutions, which often become prohibitively expensive and technically unstable when scaled to the size of the entire web.

Key Technical Advantages:

• Optimized Indexing: By fine-tuning index structures for specific tasks, Keenable avoids the massive overhead associated with scanning the entire internet for every query.
• Rapid Narrowing: The platform excels at quickly narrowing the search space, which is critical for maintaining low latency in AI agent workflows.
• Web Query Language (WQL): An upcoming proprietary feature designed to help AI systems synthesize answers by aggregating information across multiple disparate sources, even when no single source provides a complete response.

Challenging the Giants

The market for web-scale search infrastructure is currently dominated by tech titans like Google and Microsoft. However, these incumbents are increasingly restricting access to their search APIs to prevent cannibalization of their own AI products. This defensive posture has created a vacuum that startups like Keenable are eager to fill.

Styskin’s conviction in this opportunity was solidified during his tenure at Amazon, where he worked with Petri on search infrastructure for Alexa. Observing data from Cloudflare regarding the surge in AI-driven crawler traffic, he realized that the internet needed a dedicated search backbone built specifically for machine consumption.

The Innovator’s Dilemma

While acknowledging that displacing Google is an "extremely hard" task, Styskin believes the tech giant is vulnerable in the realm of agentic queries. He argues that a smaller, more agile company can innovate faster and provide a more cost-efficient solution for AI-first enterprises.

However, the path forward is not without its hurdles. Building and maintaining a search index of this magnitude is a capital-intensive endeavor. Styskin candidly admits that the costs are "painfully expensive," but the company is focused on disciplined growth and pacing its infrastructure expansion.

Looking Ahead: The End of "Ten Blue Links"

With a current team of 15 engineers spread across the United States and Europe, Keenable plans to double its headcount by the end of the year. The primary objective is to accelerate its go-to-market strategy and solidify its position as a critical utility for the AI ecosystem.

Keenable is not alone in this pursuit; competitors like Brave and Exa are also vying for a piece of the AI-search market. Furthermore, Google’s own aggressive overhaul of its search experience suggests that the industry at large recognizes that the traditional "ten blue links" model is rapidly losing relevance.

Whether the future of search belongs to the incumbents or the agile newcomers, one thing is clear: the internet is being rebuilt, not for the eyes of humans, but for the processing power of AI agents. Keenable is betting that by providing the right infrastructure today, it will become the indispensable engine of that future.