PageIndexFrom the builder

PageIndex File System: Massive-Scale Document Search

PageIndex launched the PageIndex File System, a layer above its vectorless retrieval engine that lets one index reason across millions of documents, shipping in the enterprise edition first with a cloud edition to follow. The company says the open-source project has 26,000 GitHub stars and more than 23,000 cloud users. It argues that million-document corpora often lack a folder hierarchy or have one-dimensional, unreliable labels. The new layer therefore builds virtual nodes from topic clusters and model-inferred metadata and assembles a query-specific tree without re-ingesting or re-embedding anything.

Official blogMay 3, 20269 min read