---
title: "Keenable’s $26M Seed: Web Search Infrastructure Built for AI Agents"
description: "Keenable emerged with $26M seed led by Accel (Conviction participating) to build a 100B+ document index and retrieval APIs for AI labs and agents. Gradium partnership named; other customers undisclosed. Accel has no /fund/ page here."
date: 2026-08-25T00:00:00.000Z
tags: ["2026-vc-news", "startup-funding", "venture-capital", "artificial-intelligence", "infra-cloud"]
source: https://venturecapitaltracker.com/2026-keenable-26m-seed-accel-agent-search
---

# Keenable’s $26M Seed: Web Search Infrastructure Built for AI Agents

> Keenable emerged with $26M seed led by Accel (Conviction participating) to build a 100B+ document index and retrieval APIs for AI labs and agents. Gradium partnership named; other customers undisclosed. Accel has no /fund/ page here.

**Keenable** came out of stealth with a **$26 million seed** led by **Accel** on **August 25, 2026**. Conviction Partners and angels participated. Accel has **no** `/fund/` page here — vehicle context lives on [Accel’s $3.5B early-stage funds](/2026-accel-3-5b-early-stage-funds).

**Unexpected truth:** the company already claims a **100 billion+ document** index in production at unnamed AI labs — while telling TechCrunch building it is “painfully expensive.” The scarce asset is **cost-efficient retrieval for agents**, not another ten-blue-links consumer box.

This page upgrades Keenable from roundup/entity-only coverage. Last verified **August 30, 2026**. Primary: TechCrunch.

## Five-minute decision

| If you need… | Verdict |
| --- | --- |
| What happened | **Funded.** $26M seed. Accel lead. **No valuation.** |
| What Keenable is | Independent web index + APIs (REST/MCP/CLI per entity) for agents, not humans. |
| Whether it is shipping | **Company:** production at unnamed labs; Gradium partnership live. |
| Whether to diligence | **Yes**, if agent grounding cost is your bottleneck. **No**, if you need named logos. |

**Investigate further when:** Google/Microsoft search API restrictions force labs to buy a third-party index.

**Wait or pass when:** you need Accel on `/directory` or public customer references.

## What happened

| Field | Detail |
| --- | --- |
| Company | Keenable (co-founders Andrey Styskin, Matthias Petri) |
| Round | **$26M seed** · Aug 25, 2026 |
| Lead | Accel (Zhenya Loginov) — **no** `/fund/` page |
| Other | Conviction Partners + angels |
| Index claim | **100B+** documents (company via TechCrunch) |
| Team | ~15 engineers US/Europe; plan to double by year-end |
| Named partner | Gradium (live retrieval for voice AI) |

Styskin previously led Yandex search/AI/cloud and worked Amazon search infra with Petri (TechCrunch).

## Who uses the product — and for what job

**Users:** AI labs, inference providers, and agent builders that need to ground answers in web documents at training and runtime.

**Named:** Gradium only. Other customers “several AI labs” — undisclosed.

**Job:** narrow the search space fast enough that scanning “the whole internet” for every agent query does not bankrupt the inference bill.

Upcoming: **Web Query Language** — combine partial answers across sources (company roadmap).

## Why now

- Chatbots and agents read more of each page than humans; human SERP infra is the wrong cost curve.
- Accel argues hyperscalers are shutting or bundling general search APIs to avoid cannibalization — creating a wedge for independents.
- Cloudflare-cited AI crawler growth (Styskin via TechCrunch) is the demand signal Accel underwrote.

## Why Accel — portfolio fit (judgment)

Accel’s early-stage vehicles are writing large seeds into **infra that every AI lab must rent**. Keenable sits next to retrieval/inference picks: if Fireworks-class serving is the compute layer, Keenable wants to be the **web memory** layer. Without an Accel Tracker profile, use the [Accel fund-close memo](/2026-accel-3-5b-early-stage-funds) for LP/vehicle context only.

## Keenable vs Google/Bing APIs vs Exa — what differs

| | Keenable | Google / Microsoft search APIs | Exa / Brave (named peers) |
| --- | --- | --- | --- |
| Buyer | AI labs / agents | Historically mixed; now restricted/bundled (Accel claim) | Developers / AI apps |
| Design target | Agent query volume & cost | Human SERPs | Independent retrieval |
| Disclosure | Customers mostly secret | Public products | Varies |

## When not to use this print

- **100B documents** and “production at labs” are company claims without named references.
- Do not invent `/fund/accel`.
- Not a consumer search engine story.
- Index build cost is acknowledged as extreme — runway vs crawl spend is the diligence.

## Takeaways

**Founder:** if you sell agent infra, Keenable’s buyer is the lab that lost its cheap Google API.

**Investor:** ask for cost per grounded query and retention at the unnamed labs — the 100B slide is vanity without unit economics.

**Operator:** Gradium is the only public design partner; treat everything else as stealth GTM.

**Next:** [Keenable entity](/startup/keenable) · [Accel vehicles](/2026-accel-3-5b-early-stage-funds) · [August 29–30 index](/2026-august-29-30-investment-news-restaurant-apps-trust)

**By:** [Venture Capital Tracker](https://venturecapitaltracker.com/editorial-policy)
**Last updated:** August 30, 2026

**Editorial note:** AI tools assisted with research, structure, or drafting. Venture Capital Tracker retains human editorial responsibility for factual accuracy, relevance, and source quality before publication.