Keenable is a web search API built specifically for AI agents rather than human readers, giving developers at AI labs and inference providers a way to ground model outputs in an index of more than 100 billion documents . The pitch to builders is speed and cost at web scale: a real-time search API fast enough for voice agents to answer mid-sentence, while reducing prohibitive search costs . It targets teams building retrieval-augmented chatbots, coding agents, and voice assistants that need fresh, source-grounded results without running their own crawl.
The launch matters because Keenable is stepping into a category (agent-facing search infrastructure) that already includes Exa, Brave, and Google's own AI overhaul, and it is doing so with a credible technical bet from founders who have built search at scale before. Co-founder Andrey Styskin is the former CEO of Yandex Search and a director at Amazon AGI, where he worked with co-founder Matthias Petri, previously a principal applied scientist at Amazon AGI. At Yandex, Styskin built a 200-billion-document web index. The company argues that agents read and reason over far more content per query than humans do, which changes what a good index and ranker look like.
The product is already in production. Keenable's API is used at several AI labs and inference providers during both training and runtime , and the company recently struck a partnership with voice AI company Gradium to support live information retrieval . It is coming out of stealth alongside a $26 million seed round led by Accel with participation from Conviction Partners, positioning itself as search infrastructure aimed at becoming an independent, agent-native alternative to general web search .
Comments (7)
Sign in to join the discussion.
P
Priya V.Aug 26, 2026
The 'answer isn't wrong, just very average' line is genuinely the best framing of LLM mediocrity I've read this month. Whoever wrote the tweet copy, give them the keys to the marketing budget.
T
Tomasz K.Aug 26, 2026
An AI-native index of the best human knowledge... starting with the open web. Respectfully, isn't that where the average answers came from in the first place?
T
Tomasz K.Aug 26, 2026
To be clear I want this to work, I'm just tired of 'we index the web but better' pitches that end in a wrapper.
M
Miko AdeyemiAug 26, 2026
Curious if there's an SDK or if we're all just hitting a REST endpoint and praying. Also, any plans for a public eval set so we can actually compare against the incumbents?
L
leila h.Aug 26, 2026
Every search company launches saying answers today are 'average' and every search company ends up shipping average answers. Break the curse.
M
Matteo B.Aug 26, 2026
A search engine for AI walks into a bar, orders whatever the last customer ordered. Real question: who's the paying customer, model labs or app devs?
H
Hana OzawaAug 26, 2026
Landing page loads quick but the tweet buried the actual demo below the fold. That reveal deserved a 15-second clip on top, not a link at the end.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Keenable Ai
The search engine for AI
The search engine for AI
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Keenable is a web search API built specifically for AI agents rather than human readers, giving developers at AI labs and inference providers a way to ground model outputs in an index of more than 100 billion documents . The pitch to builders is speed and cost at web scale: a real-time search API fast enough for voice agents to answer mid-sentence, while reducing prohibitive search costs . It targets teams building retrieval-augmented chatbots, coding agents, and voice assistants that need fresh, source-grounded results without running their own crawl. The launch matters because Keenable is stepping into a category (agent-facing search infrastructure) that already includes Exa, Brave, and Google's own AI overhaul, and it is doing so with a credible technical bet from founders who have built search at scale before. Co-founder Andrey Styskin is the former CEO of Yandex Search and a director at Amazon AGI, where he worked with co-founder Matthias Petri, previously a principal applied scientist at Amazon AGI. At Yandex, Styskin built a 200-billion-document web index. The company argues that agents read and reason over far more content per query than humans do, which changes what a good index and ranker look like. The product is already in production. Keenable's API is used at several AI labs and inference providers during both training and runtime , and the company recently struck a partnership with voice AI company Gradium to support live information retrieval . It is coming out of stealth alongside a $26 million seed round led by Accel with participation from Conviction Partners, positioning itself as search infrastructure aimed at becoming an independent, agent-native alternative to general web search .
Comments (7)
The 'answer isn't wrong, just very average' line is genuinely the best framing of LLM mediocrity I've read this month. Whoever wrote the tweet copy, give them the keys to the marketing budget.
An AI-native index of the best human knowledge... starting with the open web. Respectfully, isn't that where the average answers came from in the first place?
To be clear I want this to work, I'm just tired of 'we index the web but better' pitches that end in a wrapper.
Curious if there's an SDK or if we're all just hitting a REST endpoint and praying. Also, any plans for a public eval set so we can actually compare against the incumbents?
Every search company launches saying answers today are 'average' and every search company ends up shipping average answers. Break the curse.
A search engine for AI walks into a bar, orders whatever the last customer ordered. Real question: who's the paying customer, model labs or app devs?
Landing page loads quick but the tweet buried the actual demo below the fold. That reveal deserved a 15-second clip on top, not a link at the end.