Firecrawl has raised $75M in a Series B round led by Smash Ventures, the Los Angeles venture firm, with Y Combinator, Altos Ventures, Nexus Venture Partners, Freestyle and Offline Ventures joining.
The company sells web scraping infrastructure built for AI agents rather than for human readers. Its pitch is that agents can theoretically fetch anything with a normal browser, and in practice fall over on the details.
Three of those details carry the product. Firecrawl runs a search engine tuned for machines, letting developers filter candidate pages by creation date and by the kinds of files they host. A smart wait feature holds a page until every loading phase has finished, which prevents the partial reads that corrupt training and retrieval data. The platform also handles content that only appears after a user scrolls.
The round lands in a busy market. Labs are buying data at record rates, and training-set vendors such as Snorkel AI have taken large cheques this quarter to supply labelled material.
Scraping sits on contested ground. Publishers have sued over bulk collection, and the same technical tricks that make agents efficient make them harder to distinguish from bots that ignore terms of service. Firecrawl is now funded well enough to find out where that line settles.
