Firecrawl has raised $75 million in Series B funding and launched a new service called Alexandria, betting that the next bottleneck for AI agents will not simply be model intelligence but reliable access to information.
The round was led by Smash Capital, with participation from Altos Ventures, Nexus Venture Partners, Y Combinator, Freestyle and Offline Ventures.
Firecrawl said Alexandria brings official data providers, custom connectors, its own indexes and the live web into one retrieval layer.
SiliconANGLE independently reported that Firecrawl raised $75 million to expand its web-data platform for AI agents.
AI agents have a data-access problem
Large language models already contain substantial information in their training data, but agents designed to perform real-world work frequently need information that is newer, private, structured differently or inaccessible through ordinary search.
Web retrieval itself is also messy.
SiliconANGLE noted that websites can load information dynamically, reveal content only after interaction or contain difficult-to-parse files, creating problems for automated systems attempting to gather data reliably.
Firecrawl built its business around handling those tasks for developers: crawling sites, rendering pages, extracting information and returning content in formats agents can use.
Alexandria expands that idea beyond the open web.
Alexandria combines different kinds of knowledge
Firecrawl says the new service gives agents access to its Research, Developer and Government indexes, external data providers and specialized tools through a common interface.
The company reported that in its own evaluation, agents using Alexandria scored 21% higher on answer quality than agents using built-in web tools across 845 tasks, using the same model and prompts with blind AI judging.
The company is also taking an unusual approach to content acquisition.
Firecrawl says part of the Series B will be spent buying knowledge from publishers, researchers and data providers, rather than assuming all useful information should be freely scraped from the internet.
Dealroom reported that Alexandria is designed to combine scientific research, code, public documents and real-time information for models and agents.
Retrieval is becoming infrastructure for autonomous software
The funding round points to an important shift in the AI stack.
During the first phase of generative AI, much of the investment concentrated on foundation models. As agents attempt longer and more complicated tasks, information retrieval becomes increasingly important.
A capable model working from stale, incomplete or poorly parsed data can still make poor decisions.
That gives companies building search, indexing, crawling and data-connectivity layers a growing role.
Firecrawl says more than 1.25 million developers have signed up for its technology and that more than 150,000 companies use its services.
For developers, Alexandria’s pitch is essentially that agents should not need a different integration for every source they want to consult.
That could make data access a standardized part of the agent stack rather than something engineering teams repeatedly build themselves.
The larger competition in AI may therefore be shifting.
Models still matter enormously. But as autonomous systems become more capable, whoever controls the infrastructure that lets them find trustworthy information at the moment they need it could become just as important.
