Diffbot

Diffbot automatically extracts structured data from web pages using computer vision and ML, and maintains a Knowledge Graph of over 10 billion entities and a trillion-plus facts. It is a developer-first platform with extraction APIs, Crawlbot, and Enhance, offering a free credit tier and credit-based paid plans. It is a standout for web-scale structured data and entity intelligence, but is technical and can be costly at scale.

Diffbot's extraction APIs (for articles, products, discussions, and more) automatically parse arbitrary web pages into structured JSON using computer vision and ML, without site-specific rules. Its Crawlbot can crawl entire sites, and Diffbot Enhance can append firmographic and entity data to records. The result is that developers can extract clean structured data from pages they have never configured for. The flagship differentiator is the Diffbot Knowledge Graph, a machine-constructed graph reportedly spanning over 10 billion entities and more than a trillion structured facts, refreshed continuously from crawling the public web. It powers use cases in market intelligence, lead enrichment, news monitoring, and research, queried via API. Diffbot was founded in 2008 (formally around 2012) by Mike Tung and is based in the Bay Area. It raised a relatively modest amount of venture funding, including a 2016 Series A led by Tencent and Felicis, and has operated as a focused, technically deep company. It is developer-oriented and priced with a credit model that can suit both experimentation and heavy production use.