Keenable Raises $26M Seed Led by Accel: 100-Billion-Document Search Index Built for AI Agents with MCP Integration
On August 25, 2026, AI infrastructure startup Keenable announced its exit from stealth mode, simultaneously disclosing a $26 million seed round led by Accel. Keenable's core product is an independent web index containing over 100 billion documents, designed specifically for AI agents rather than human users, aimed at addressing AI agents' urgent need for reliable, low-latency web data when executing tasks.
Why Do AI Agents Need Dedicated Search Infrastructure?
Traditional search engines (like Google and Bing) are designed for human users — they optimize for the "ten blue links" presentation format, suitable for humans with limited attention spans to quickly browse. However, AI agents' search needs differ fundamentally from humans:
Scale Difference: AI agents may issue hundreds or even thousands of search requests in a single task, while humans typically view only a few results per search.
Content Requirements: AI agents need complete page content and structured data, not summary snippets.
Latency Requirements: AI agent workflows are extremely sensitive to latency — any search delay directly impacts overall task completion time.
Verifiability: AI agents need the ability to trace information sources to reduce hallucinations and improve response credibility.
Cost Efficiency: In high-frequency search scenarios, traditional search API costs can accumulate rapidly, requiring more cost-effective solutions.
Keenable's Technical Architecture
Keenable has built an independent web index containing over 100 billion documents, a scale that makes it one of the world's largest AI agent-dedicated search indexes.
Core Technical Components
Search API: Provides a low-latency interface delivering ranked results and full-page content retrieval for direct use by AI models. Unlike traditional search APIs, Keenable's API returns complete page content rather than just summaries.
MCP Integration: Keenable provides a Model Context Protocol (MCP) server, allowing AI agents like Cursor and Claude Code to interface with the index without requiring custom glue code. This feature significantly lowers the technical barrier for AI agents to integrate search capabilities.
Web Query Language (Coming Soon): A feature allowing AI systems to synthesize information across multiple web sources, even when a single page doesn't contain a complete answer.
Historical Queries: The platform supports point-in-time retrieval, allowing agents to access web data as it existed at specific historical moments — particularly useful for tasks requiring knowledge of past events or trends.
Founding Team Background
Keenable was co-founded by two founders with deep backgrounds in search and AI:
Andrey Styskin: Previously led search, AI, and cloud divisions at Yandex, and worked on search infrastructure for Amazon's Alexa. His deep understanding of large-scale search systems is central to Keenable's technical architecture.
Matthias Petri: A German AI scientist with extensive experience in Amazon's AGI retrieval systems.
Funding Details and Investors
The $26 million seed round was led by Accel, with participation from Conviction Partners, Brightwing Capital, ScOp Venture Capital, and various angel investors from Google and Amazon. The funding actually closed on November 5, 2025, but the company chose to publicly disclose it on August 25, 2026.
Business Model and Pricing
Keenable targets AI labs and inference providers, positioning itself as a "picks and shovels" provider that avoids direct competition with its customers. The pricing structure includes:
- Free Tier: For evaluation use
- Pay-as-you-go: $4 per 1,000 requests
- Enterprise Pricing: $1 per 1,000 requests for high-volume users
This pricing strategy makes Keenable more cost-competitive than traditional search APIs in high-frequency search scenarios.
Competitive Landscape
Keenable operates in an increasingly competitive market alongside Exa, Brave, and Parallel Web Systems. Keenable's differentiation lies in:
- Scale: A 100-billion-document index that leads in the AI agent-dedicated search space
- Native MCP Support: Seamless integration with mainstream AI agent frameworks
- Historical Query Capability: Support for point-in-time data retrieval
- Founding Team's Search Infrastructure Expertise: Deep backgrounds from Yandex and Amazon
Asia-Pacific AI Agent Search Demand
The AI agent market in the Asia-Pacific region is developing rapidly. According to a Temporal report, frequent AI agent usage among engineers increased by 70.8%, with 80.8% of surveyed professionals now using agents daily.
For AI developers and enterprises in the Asia-Pacific region, Keenable's launch provides an important infrastructure option, particularly in the following scenarios:
Multilingual Search: AI agents in the Asia-Pacific region need to handle search requests in Chinese, Japanese, Korean, Hindi, and other languages. Whether Keenable's 100-billion-document index covers sufficient Asia-Pacific language content will be key to its success in the APAC market.
Compliance Requirements: Different Asia-Pacific countries have varying requirements for data localization and privacy protection. Keenable will need to address these regulatory challenges when expanding into the APAC market.
Broader Trends in AI Agent Search Infrastructure
Keenable's emergence reflects the rapid maturation of the AI agent infrastructure market. AWS has brought web search on Amazon Bedrock AgentCore to general availability (GA), allowing agents to fetch cited web data within secure cloud boundaries. Google's A2A protocol has joined the Agentic AI Foundation (AAIF), aligning with Anthropic's MCP under neutral Linux Foundation governance.
These developments collectively indicate that AI agent infrastructure is moving from fragmented solutions toward standardization and interoperability.
Conclusion
Keenable's $26 million seed round and 100-billion-document index represent important progress in the AI agent infrastructure market. By focusing on AI agents' specific needs — low latency, complete content, MCP integration, and historical queries — Keenable has found a clear positioning in a rapidly growing market. For AI developers and enterprises in the Asia-Pacific region, this is an emerging infrastructure provider worth watching.
