Keenable, a San Francisco startup, has raised $26 million in a seed funding round to develop a search engine specifically designed for artificial intelligence. Unlike traditional search engines like Google, which were built for humans where entering a few words and clicking the first result is sufficient, an AI agent may require analyzing thousands of pages to answer a single question.
Accel led this round, with participation from Conviction Partners and investors from Amazon, Google, and Databricks. The company was founded by Andrey Styskin and Matthias Petri in 2025.
Both founders spent twenty years working at major search engines before observing how AI is transforming the field. As early as 2024, at the USC-Amazon symposium, Styskin warned that scanning the entire internet for large language models (LLMs) would become too expensive without a new index design. Petri's research focused on low-latency retrieval and compression—problems currently faced by AI agents.
After leaving Amazon, they decided to 'separate' search from large tech companies and offer it as infrastructure accessible to any AI company. Keenable has already amassed a massive library of internet data for AI use, boasting a web index of over one hundred billion documents. AI laboratories are already using their API for both training and live searching. Furthermore, the company has partnered with the voice AI company Gradium.
Keenable is developing a 'Web Query Language' that will allow AI to synthesize answers from multiple websites. They also provide a 'point-in-time retrieval' feature, enabling AI to view the appearance of a webpage on a specific date. Service costs start at $1 per 1000 API queries.
Keenable's technology allows for rapid narrowing of searches based on AI prompts, which, according to the founders, makes web search approximately 10 times cheaper for AI workloads. This is becoming relevant as Google and Microsoft restrict access to their search APIs, leaving AI companies searching for new solutions. Keenable is not aimed at standard Google searches but at AI agents performing tasks such as market mapping, price monitoring, and lead research—any work requiring fast reading and summarization of large volumes of web pages.
With the new capital, Keenable plans to double its staff by the end of the year, growing its current team of about fifteen people to thirty by the end of 2026. The main product goal is the 'Web Query Language,' and the ultimate goal is to make full-scale web search significantly more accessible so that every AI application can afford to stay up-to-date.