A new startup called Keenable emerged from stealth on August 25 with a $26 million seed round and a pointed premise: the web wasn’t built for AI agents, and the search infrastructure underpinning it wasn’t either.

Accel led the round, with Conviction Partners and a group of business angels also participating. The company was co-founded by Andrey Styskin, who previously led search, AI, and cloud at Yandex, and Matthias Petri, a German AI scientist. Together, they’re building something the existing internet stack doesn’t have: a search index optimised for machines, not people.

What Keenable Actually Does

Traditional web indexes were architected for browsers and human-scale queries. A person types ten words and gets ten blue links. AI agents work differently. They make hundreds of calls, retrieve full documents, and need to ground their responses in source material that is accurate, current, and citable.

Keenable is building a web index of more than 100 billion documents, and its API is already running in production at several AI labs and inference providers. The company’s focus is on what it calls “grounding”: connecting AI chatbot responses to actual source documents, not summaries of summaries.

The distinction matters more than it might sound. When an AI agent hallucinates, it is almost always because it was operating without a reliable retrieval layer. The better the index, the fewer the errors.

Why This Matters Right Now

The timing lines up with a structural shift in how AI is being deployed. AI agents no longer just answer questions. They take actions, retrieve information, synthesise reports, and interact with business systems autonomously. The infrastructure those agents rely on is becoming a serious investment category.

Web search was already considered solved for consumers. For agents, it is still a fresh problem. Keenable’s pitch is that agents need different things: larger context windows, more recent data, structured retrieval, and the ability to pull complete documents rather than page previews.

The $26M seed at exit from stealth signals that the market agrees. Accel and Conviction don’t write that size check without some level of production usage already in place, and Keenable confirmed its API is live with AI labs and inference providers.

The Business Angle

For companies building AI agents, whether through internal builds or services like Omni Ops, the quality of the retrieval layer directly determines the quality of outputs. An agent summarising a supplier contract, analysing competitor pricing, or researching a client’s industry needs fresh, verifiable source data to work from.

Companies like Keenable are building the plumbing that makes agents more reliable in practice. The better this infrastructure, the fewer cases where an agent confidently gives a wrong answer because it was working from stale or missing data.

This is also part of a broader pattern. A growing number of AI infrastructure companies are raising at the seed and Series A stage specifically to solve problems that weren’t visible until agents started running at scale: memory, orchestration, identity, retrieval, cost management. Keenable is squarely in that stack.

What This Means for Business

If you are evaluating or building AI agents for business use, the retrieval layer is worth interrogating. Ask whether the agent has access to current web data, whether it cites sources, and whether it can distinguish between a document from last week and one from three years ago.

Keenable’s emergence suggests the market is taking this problem seriously enough to fund it at scale. It also signals that the AI agent ecosystem is maturing past the model layer and into the infrastructure below it, where durable value gets built.

The company hasn’t announced public pricing or a general API product yet, but its production deployment with AI labs suggests a commercial offering is in the works.

---

Enterprise DNA helps businesses understand and adopt AI. Learn more about how Omni Ops can put AI agents to work in your business.