Keenable: The Independent Web Search Index for AI Agents
Keenable is a web search company built for AI agents. It crawls and indexes the web itself, more than 100 billion documents according to press coverage, and sells fast search and page fetching to agents, AI labs and inference providers. Its team comes from two of the world’s biggest search operations: its CEO previously ran search and ads at Yandex, and its chief scientist built the web grounding service behind Amazon’s Alexa. Keenable came out of stealth in August 2026 with a $26 million seed round backed by Accel and Conviction Partners, and released Keenable 1.0 on November 12, 2026.
The quick answer: Keenable is the newest serious entrant among search APIs for AI, with an experienced search team, its own index and one of the most generous free tiers around: 100,000 requests a month. Its unusual features include searching the web as it was at a past date and a query language that computes answers across thousands of pages. It’s very young, and its customers aren’t named yet. Everything below comes from Keenable’s site, pricing, documentation and launch post, plus press coverage.
Keenable at a glance
Checked against Keenable’s site, pricing page, developer documentation and Keenable 1.0 launch post on November 13, 2026.
| Keenable | |
|---|---|
| What it is | Web search and page fetch for AI agents, from Keenable’s own index |
| Index | More than 100 billion documents, per press reports |
| Speed | Under 250 milliseconds at the 95th percentile in US East, per Keenable |
| Free tier | 100,000 requests a month; limited use with no key at all |
| Paid pricing | $4 per 1,000 requests; $1 per 1,000 on dedicated capacity |
| Access | MCP server, CLI, and SDKs for TypeScript, Python and Go |
| Company | San Francisco; founded 2025 by Andrey Styskin (CEO) and Matthias Petri (chief scientist) |
What Keenable does
Keenable’s developer documentation describes it simply: “Web search and page fetch for AI agents, on Keenable’s own index.” An agent sends a query and gets up to 50 results, each with a title, URL, description, a longer snippet, the publish date and when Keenable last indexed the page. It can also fetch a page’s content, either Keenable’s indexed copy or a live fetch from the source.
Its homepage describes three products built on the same index:
- Search API for agents, available over REST, MCP and a CLI.
- Point-in-time search, which searches the web as it existed at a past moment. Add a time to the query and “both the corpus and the ranking rewind.” That’s useful for testing agents and training models without leaking future information.
- Web Query Language, which turns thousands of web pages into a table of values and computes an answer across its rows, with a source on every row. Keenable’s example tracks researcher moves between AI labs, each with a link to where it was reported.
What Keenable 1.0 added
Keenable’s launch post on November 12, 2026 says the index “learns from every agent query and gets better with every iteration,” and that millions of agent queries since the August launch showed what agents need: authoritative sources, fresh information and context that doesn’t waste tokens. Version 1.0 added two features:
- Knowledge cards, frequently refreshed summaries of key facts on the topics agents search for most, so they don’t have to rebuild the same answer repeatedly.
- Dynamic snippets, which read whole documents and pull out just the context an agent needs, which Keenable says cuts token use by 50% on research tasks.
Keenable says 1.0 reached first place on the Artificial Analysis Search Index, across the main AI web search benchmarks (BrowseComp, DeepSearchQA and AA-Omniscience) and on the Vals Web Search Index. It also runs its own open-source benchmark, NEEDLE, on fresh queries daily. Rankings in this category move quickly and every vendor highlights the ones it leads, so test on your own queries.
Built for agents to sign up and pay
Like several newer rivals, Keenable is designed for agents to adopt with little human help. Search and fetch work with no account or key, limited to 1,000 requests an hour per IP address. One pasted prompt has a coding agent install the SDK, run a test search and show the cited results. Agents can even pay per call without signing up, at $0.01 per search or fetch in USDC stablecoin, or buy credits through an agent storefront with their human’s approval.
Pricing
Keenable’s pricing is simple:
| Plan | Price | Notes |
|---|---|---|
| Free | $0 | 100,000 requests a month, no card required |
| Pay as you go | $4 per 1,000 requests | Cloud only; credits never expire |
| Dedicated capacity | $1 per 1,000 requests at 100+ requests per second | Cloud or on-premises, zero data retention available |
A search is billed as one request whether it returns 1 result or 50. When free requests run out, calls stop until you buy credits or the month resets, unless you turn on auto top-up. Keenable also offers $5,000 in credits for startups and 100,000 credits a month for academic research.
The team and funding
Keenable’s co-founder and CEO, Andrey Styskin, previously led search and ads at Yandex, the Russian search engine, and later worked on web infrastructure at Amazon’s AGI group. Co-founder and chief scientist Matthias Petri was a principal applied scientist at Amazon AGI, where he built the web grounding service behind Alexa. According to SiliconANGLE, Styskin saw the need for a new kind of search while working with Petri on web infrastructure for AI at Amazon.
Keenable launched publicly on August 25, 2026 with a $26 million seed round backed by Accel and Conviction Partners, with Google and Amazon executives among its angel investors. At the time it had 15 employees and planned to double that. It says it’s in production at several AI labs and inference providers but hasn’t named them. Styskin explained the technical bet to TechCrunch: “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.”
How Keenable compares
Keenable competes with Exa, Parallel, the Brave Search API and Tavily, and its fetch overlaps with scrapers like Firecrawl and Context.dev, whose daily benchmark includes Keenable. Like Exa, Brave and Parallel, it runs its own index rather than relying on Google or Bing. Its distinguishing traits are its free tier of 100,000 requests a month, point-in-time search, its Web Query Language and a team with years of experience running consumer search at scale. Its main drawback is age: Exa, Brave and Tavily have longer track records and named customers.
If you’d rather not wire up search yourself, Carly searches the web as part of its work and connects to thousands of apps, so from ChatGPT, Claude or a text message you can ask it to check the latest on a client and add a note to your CRM.
Who Keenable suits
Keenable is a good fit if:
- You want an independent search index with a large free tier to start.
- Your agents make many fast searches and latency matters.
- You need to search the web as it was at a past date, for evaluation or training.
- You’re an AI lab or inference provider that needs dedicated or on-premises search.
Look elsewhere if:
- You want a provider with a long track record and named customers.
- You need deep, multi-step research jobs or structured data on people and companies.
- You want a finished assistant rather than an API.
FAQ
What is Keenable?
Keenable is a web search company for AI agents. It runs its own index of more than 100 billion documents and offers search, page fetch and point-in-time search through an API, an MCP server and SDKs.
Is Keenable free?
Keenable gives every organization 100,000 free requests a month with no card required, and search and fetch work without a key at a limited rate. Beyond that, it costs $4 per 1,000 requests.
Who founded Keenable?
Andrey Styskin, formerly head of search and ads at Yandex, and Matthias Petri, formerly of Amazon AGI, founded Keenable in 2025.
Does Keenable have an MCP server?
Yes. Keenable offers an MCP endpoint that gives MCP-compatible AI tools its search and fetch, alongside a CLI and SDKs.
How much funding has Keenable raised?
Keenable launched in August 2026 with a $26 million seed round backed by Accel and Conviction Partners.
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
