All Labels

  • Published on
    We propose a practical acquisition function for prompt/completion pairs based on the predictive entropy of the language model and a measure of certainty of the implicit preference model optimized by DPO.
  • Published on
    We introduce Mafin2.5, which is built based on PageIndex, with a 98.7% accuracy rate on the finance industry question-answering benchmark.
  • Published on
    The same bet behind both PageIndex and Claude Code: skip the vector DB and let the LLM itself drive retrieval. We explore the rise of agentic retrieval over vector indexing and PageIndex's agentic, vectorless RAG framework. Where Claude Code retrieves over a codebase with simple bash tools instead of a vector database, PageIndex gives long documents an in-context tree index that an LLM agent navigates by reasoning — no chunking, embeddings, or vector store.
  • Published on
    We argue that context blindness — the inability of vector-based retrieval to condition on full conversational and reasoning context — is a fundamental limitation of vector RAG, and outline a paradigm shift from semantic similarity to context-dependent relevance classification. In this view, retrieval becomes a relevance decision made by an LLM with full context, scaled efficiently through hierarchical tree search.
  • Published on
    Why discard document information before knowing what the user will ask? We examine the information bottleneck in OCR-first pipelines, then show how to design retrieval, QA, and extraction around task-guided reading of the original pages.
  • Published on
    We built what Andrej Karpathy described, and solved the hard part. OpenKB is an open-source CLI that compiles raw documents into a structured, interlinked wiki, powered by PageIndex for long PDFs.
  • Published on
    Page Extract API turns any PDF into the JSON you describe: write the fields you want as a JSON Schema, and get them back from the document, every row of every table included. It is the most accurate method on ExtractBench, and its standard tier still beats every other method at $0.03 a page, 60% less than the next most accurate.
  • Published on
    Page Search API finds the pages that answer each question in a long document, with a relevance score for each, so your LLM reads only those pages instead of the whole document. It finds 42% more of the pages that answer than the best vector search in its top 5, searches a new 160-page document in 1.1 seconds with no indexing, reads charts and scans without OCR, and cuts the tokens you send your LLM by 97%.
  • Published on
    The new PageIndex Agent finds answers in long documents the way an expert does: it picks the right documents, follows each one's structure to the right section, narrows it to the pages that answer, and reads them visually, charts and tables included. Every answer has a line-level citation. No OCR, no embeddings, no vector database.
  • Published on
    Experience the power of reasoning-based RAG with PageIndex Chat - our new conversational interface for intelligent document understanding.
  • Published on
    PageIndex File System is a file-level tree layer that sits above your documents and scales the same PageIndex tree search from a single document to millions of documents in one index. It synthesizes a semantic hierarchy with virtual nodes when no usable folder structure exists, builds the tree on demand for each query, and adapts how it searches each node to stay efficient at scale.
  • Published on
    PageIndex Flash runs entirely on your own machine and is designed for text-based PDFs. It builds a hierarchical tree index from a PDF's own layout, indexing a 1,000-page document in minutes for about a dollar. PageIndex Flash is available now in the PageIndex SDK, with your own LLM and API key, page-level citations, no chunking and no vector database. Flash is now also available on PageIndex Cloud.
  • Published on
    PageIndex is a vectorless, reasoning-based retrieval framework that simulates how human experts extract knowledge from complex documents. Instead of relying on vector similarity search, it builds a tree-structured index from documents and enables LLMs to perform agentic reasoning over that structure for context-aware retrieval. The retrieval process is traceable and interpretable, and requires no vector DBs or chunking.
  • Published on
    We benchmarked PageIndex Chat against ChatGPT 5.1 on real-world long documents. PageIndex achieved 100% accuracy compared to ChatGPT 5.1's 59-82%, with faster response times and page-level traceability.
  • Published on
    How PageIndex’s vectorless, reasoning-based RAG overcomes the challenges of traditional vector RAG in long, complex technical manuals.