Agentic web search infrastructure startup Keenable raises $26M

AI

Artificial intelligence startup Keenable.ai Inc. said today it’s launching with $26 million in funding to try to revamp a web search infrastructure that was built for humans, not the billions of autonomous agents that many believe will soon dominate the internet.

The seed round was backed by investors including Accel and Conviction Partners and saw participation from a number of unnamed Google LLC and Amazon.com Inc. executives as angels.

Keenable has developed an independent web search index that spans more than 100 billion documents. Through its index, it optimizes search for the type of low-latency, high-frequency queries conducted by AI agents, serving use cases including market mapping, pricing monitoring and lead enrichment.

The startup has developed application programming interfaces for each of these use cases that support straightforward, natural language searches as well as cleaned content fetching. It’s able to delivery “model-ready inputs” with minimal delay. It’s priced at $1 per 1,000 API requests, aimed at “frontier-scale” users, the company said. Keenable also supports point-in-time historical queries, which means models can search the internet as it existed at a specific point in time, instead of accessing it as it exists right now.

Keenable’s co-founders know a thing or two about how to search the internet. Chief Executive Andrey Styskin (pictured, right) previously worked at Yandex N.V., the Russian-language internet search giant, while Chief Scientist Matthias Petri (left) comes from Amazon AGI, where he was focused on building internet retrieval systems.

In an interview with TechCrunch, Styskin explained that Keenable’s product is very different from the enterprise-scale search systems currently used by AI models. “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,” he said. “That’s why you need to innovate on how you can narrow the search space based on your query very fast. This is what we are bringing to the table.”

Styskin said he first realized the need for a better way to search the web during his own stint at Amazon, where he worked closely with Petri on building the web search infrastructure for AI applications such as Alexa. He quickly concluded that there was an urgent need for a new kind of web infrastructure designed with AI agents in mind.

To that end, Keenable is developing both its own proprietary information retrieval system and a new product called Web Query Language, which helps AI systems to answer questions by pulling data from multiple web sources, even when none of them contain all of the information required to piece together the full answer.

The startup claims that the quality of its search results is already superior to agentic search rivals such as Tavily, Exa Labs Inc. and Perplexity AI Inc. in terms of its ability to pull up relevant, clean information that’s optimized for AI agent consumption.

Styskin told TechCrunch that there’s a massive opportunity for agentic search infrastructure. Right now, he said, most AI agents are still using traditional search APIs or else they have built their own web crawlers from scratch. Keenable offers a dedicated index that handles all of this, so developers can focus their efforts on developing their agents exclusively, rather than a search stack.

Keenable’s customers include a handful of AI labs that are already using its APIs in production, but it didn’t name any of them. Armed with today’s funding, it plans to double its headcount from the 15 employees it has now in order to accelerate its go-to-market strategy.

Photo: Keenable