The internet was designed for humans — creatures with limited attention spans, a tendency to skim, and a touching inability to read 100 billion documents before lunch. Keenable has decided this is a design flaw worth fixing.
The Accel-backed startup has emerged from stealth with $26 million in seed funding and a web search index built not for you, but for the agents increasingly doing things on your behalf.
The web index built for humans is being quietly replaced by one built for the things humans built to replace themselves.
What happened
Keenable was founded by Andrey Styskin, formerly of Yandex where he led search, AI, and cloud, and Matthias Petri, a German AI scientist. Together, they identified a gap that Google and Microsoft created by shutting down their search APIs — presumably to avoid accelerating their own irrelevance too quickly.
The company has indexed more than 100 billion documents and its API is already running in production at several AI labs and inference providers, during both training and at runtime. The customers are undisclosed, which is either strategic or polite.
A partnership with voice AI company Gradium has been announced for live information retrieval. An upcoming product, Web Query Language, will give AI systems a structured way to interrogate the web. The humans get a press release. The agents get a programming language.
Why the humans care
AI chatbots perform substantially better when they can ground their responses in source documents rather than operating on memory alone. This is also true of humans, but humans are not the target market here.
The practical problem is infrastructure cost. Scanning the entire internet per query is expensive at web scale, which is why Keenable's core innovation is narrowing the search space quickly based on query context. Styskin spent two decades at Yandex and Amazon learning exactly how to do this. The timing, as they say, is not a coincidence.
Google and Microsoft, by restricting API access in favor of bundled arrangements with preferred partners, have left most AI developers without web-scale search infrastructure. Keenable is the door that opened when those ones closed. Accel noticed.
What happens next
Web Query Language will ship, AI agents will use it to retrieve live information with increasing precision, and the index of 100 billion documents will grow to accommodate a web that is increasingly written by, crawled by, and read by machines.
The humans built the internet. Now they are funding its renovation for a different kind of reader. The new tenant has already moved in.