The web data API for AI agents — 1.5M+ users, now building Alexandria, the knowledge library that pays data providers
HQ: San Francisco, California, United States | Founded: 2022 | Employees: Not publicly disclosed (small team) | Stage: Series B | Website: https://www.firecrawl.dev
Firecrawl started as a side-effect of a different product. The founders — Caleb Peffer, Nicolas Silberstein Camara, and Eric Ciarla — built Mendable, an AI chat product for documentation that went through Y Combinator's S22 batch and was used by companies like Snapchat, Coinbase, and MongoDB. The hardest part of that stack wasn't the model; it was getting clean, reliable information out of web pages. They spun that capability into Firecrawl: hand it a URL and it handles crawling, rendering (including JavaScript-heavy pages), parsing, and cleanup into LLM-ready output.
Adoption followed the pattern of every good developer tool: open-source SDKs, generous free tiers, and word-of-mouth among teams building AI products. The company reported 1.5M+ users, with usage scaling as AI agents went mainstream — agents don't just need one page, they need to crawl sites, follow links, and extract structured data at scale. A $14.5M Series A closed in 2025.
The September 2026 Series B ($75M led by Smash Capital, with Altos Ventures, Nexus Venture Partners, Y Combinator, Freestyle, and Offline Ventures participating) arrived alongside Alexandria, Firecrawl's biggest product swing. Alexandria gives agents one way to discover sources, understand what they hold, and pull from them — the live web plus official data providers plus Firecrawl's own indexes (a Research Index of tens of millions of scientific abstracts, a Developer Index of documentation/READMEs/issues/merged PRs, and a Government Index of laws and regulations).
Two things make Alexandria more than an aggregator. First, measured impact: across 845 tasks with the same model and prompts, agents using Alexandria scored 21% higher on answer quality than agents using built-in web tools, judged blind. Second, the economic model: Firecrawl pays official data providers through agreements — most notably Wikimedia Enterprise, with millions of Wikipedia requests flowing through monthly — and plans a self-service system so individuals and creators can earn when agents use their knowledge. The company frames this as 'the library for superintelligence' — and, notably, as a market-based answer to the content-industry backlash against AI scraping.
Firecrawl won the unglamorous but universal problem of AI — getting clean data off the web — and rode it to 1.5M users and a $75M Series B. Its next act, Alexandria, is a bolder bet: a knowledge library for AI agents with curated indexes and a payment model for the people who create the world's data.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki