Top all-time · #11024
PageIndex
Company blog · pageindex.ai/blogaidata
PageIndex is the company blog at pageindex.ai, writing about ai and data. Since 2025, 7 different Hacker News readers have linked it 17 times, most often “Do we still need OCR? An implementation of a pure vision-based agent”. It is still writing: the last post was on 2026-08-26, and you can follow it by feed.
Too few sampled comments (1) to say how often HN recommends it.?
Why HN links it
HN readers link PageIndex, a "vectorless RAG" project that retrieves without vector embeddings. The one comment discusses its benchmark claims, which the commenter finds lacking.
Summarised by Claude from a sample of 1 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
PageIndex is a vectorless, reasoning-based RAG engine that mirrors how humans read documents. Achieve 98.7% accuracy on FinanceBench with traceable, explainable retrieval.
Most-linked posts?
- Do we still need OCR? An implementation of a pure vision-based agent2 people · 4 mentions · first linked 2025-10-27
What Hacker News says?
Edited version: We try to solve a similar problem to put long documents in context. We built an MCP for Claude to allow you to put long PDFs in your context window that go beyond the context limits: https://pageindex.ai/mcp .
Comment linking /mcp
on “Managing context on the Claude Developer Platform”Submitted Show HN: PageIndex OCR – The First Long-Context OCR Model as a story · 10 points
HN discussionA suspicious lack of any performance metrics on the many standard RAG/QA benchmarks out there, except for their highly fine-tuned and dataset-specific MAFIN2.5 system. I would love the see this approach vs. a similarly well-tuned structured hybrid retriever (vector similarity + text matching) which is the common way of building domain-specific RAG. The FinanceBench GPT4o+Search system never mentions what the retriev…
Comment linking /blog/Mafin2.5
on “Show HN: PageIndex – Vectorless RAG”Submitted Do we still need OCR? An implementation of a pure vision-based agent as a story · 7 points
HN discussionWe actually don't need OCR: https://pageindex.ai/blog/do-we-need-ocr
Comment linking /blog/do-we-need-ocr
on “Should LLMs just treat text content as an image?”Submitted PageIndex Chat – Human-Like Long Document AI Analyst as a story · 6 points
HN discussionHi, thanks for your inspiring questions. 1. What happens when the TOC is too long? -- This is why we choose the tree structure. If the ToC is too long, it will do a hierarchy search, which means search over the father level nodes first and then select one node, and then search its child nodes. 2. How does the index handle near misses, and how do you disambiguate between close titles? For each node, we generate a des…
Comment linking /mcp
on “Show HN: A Vectorless LLM-Native Document Index Method”Submitted Show HN: Long PDF Reader MCP as a story · 5 points
HN discussionCheck out this MCP: https://pageindex.ai/mcp , which allows you to chat with any long PDFs (hundreds of pages) beyond the context limit of Claude or ChatGPT.
Comment linking /mcp
on “Show HN: I built an AI in 3 days at 16 y/o that lets you chat with your PDFs”Submitted Do we still need OCR when we can build a pure vision-based AI agent as a story · 4 points
HN discussionSubmitted Do We Still Need OCR? as a story · 4 points
HN discussionSubmitted PageIndex Chat – Human-Like Long Document AI Analyst as a story · 4 points
HN discussionSubmitted Human-Like Document AI as a story · 2 points
HN discussionSubmitted A file-level tree that lets an LLM reason over a document corpus as a story · 1 points
HN discussionSubmitted PageIndex: Next Generation Vectorless, Reasoning-Based RAG as a story · 1 points
HN discussionSubmitted Reasoning-based RAG for long document question answering as a story · 1 points
HN discussionSubmitted Show HN: PageIndex MCP – Chat with Long PDFs on Claude or Cursor as a story · 1 points
HN discussion