{"slug": "introducing-the-new-pageindex-agent", "title": "Introducing the New PageIndex Agent", "summary": "VectifyAI launched the PageIndex Agent, a vectorless, OCR-free multimodal RAG product rebuilt and renamed from the former PageIndex SDK and Cloud, which returns answers with line-level citations. On VectifyAI's benchmark of questions about 10-K filings, the underlying Page Search engine finds 99.3% of answering pages in its top 20 results, versus 90.3% for the best embedding model tested (Voyage 4 Large), and searches a new 160-page document in 1.1 seconds with no indexing or preprocessing while returning about 97% fewer tokens than the whole document. The agent navigates a document's section tree and then narrows to pages, reading charts, tables, scanned pages and drawings visually, and each citation carries the page, block ID and bounding box; it sits alongside the Page Search API and Page Extract API as one of three PageIndex products.", "body_md": "Today we are introducing the new **PageIndex Agent**: vectorless, OCR-free, multimodal RAG for complex documents, with line-level citations for every answer.\n\nPageIndex Agent is what we used to call the PageIndex SDK and Cloud, rebuilt and renamed. It is now one of three PageIndex products, alongside [Page Search API](https://pageindex.ai/search) and [Page Extract API](https://pageindex.ai/extract), and it is the one to use when you want **answers**: you bring your documents and your own LLM, and PageIndex finds the evidence and cites it.\n\n## \n\nVector RAG splits documents into chunks and retrieves the ones whose embeddings sit closest to the question. Closeness in meaning is not the same as relevance, and in long financial filings, contracts and manuals the passage that answers often shares few words with the question.\n\nPageIndex Agent searches the way an expert reader does, in four steps:\n\n1. **Find the relevant documents.** Keep documents in collections tagged with your own metadata, or in folders, and each question goes only to the documents that match.\n2. **Find the relevant sections.** PageIndex builds a tree of each document's sections, like a table of contents, and the agent navigates it as a person would: straight to the sections that matter, each traced to its pages.\n3. **Find the relevant pages.** Page Search narrows each section to the few pages that answer, reading text and images alike.\n4. **Find it on the page.** The agent reads those pages visually and outlines the evidence it used, down to the line.\n\nIt never reads the whole document, so it spends far fewer tokens, and every step can be checked.\n\n## \n\n### \n\nStep 3 is new. Once the tree has picked a section, [Page Search](https://pageindex.ai/search) scores every page in it against the question and keeps only the pages likely to answer. The agent then reads a handful of pages instead of a whole section.\n\nPage Search is the same engine we sell as its own API, and on our [benchmark](https://github.com/VectifyAI/page-search-benchmark) of questions about 10-K filings it finds **99.3%** of the pages that answer in its top 20 results, against 90.3% for the best embedding model we tested (Voyage 4 Large). It searches a new 160-page document in **1.1 seconds**, with no indexing or preprocessing, and the pages it returns are about **97% fewer tokens** than the whole document.\n\n### \n\nThe new PageIndex Agent reads pages as they are, not as text pulled out of them. Charts, tables, scanned pages and engineering drawings are read visually, so a value that lives in a chart or a drawing is as findable as one in running text, and nothing is lost to an OCR step first.\n\nCitations follow the same idea. Each one carries the page and the block it came from, with its position on the page, so you can outline the evidence on the page rather than only name a page number:\n\n```\ncitations[0]\n# {'document': 'report.pdf', 'page': 12, 'block_id': 'p12_text_3',\n#  'bbox': [78, 24, 293, 44], 'block_type': 'text', 'text': '…'}\n```\n\n### \n\nDocuments can now be organized in two ways, whichever fits your project:\n\n- **Collections with metadata** : group documents and tag each with your own keys, such as company or year, so a question goes only to the documents that match.\n- **File system with folders** : keep documents in folders and subfolders, and ask within a folder.\n\nEither way, step 1 of the search is the organization you already have. See [Collections with metadata](https://pageindex.ai/docs/documents/organize) and [File system with folders](https://pageindex.ai/docs/documents/folders).\n\n### \n\nPageIndex now comes as three products built on the same document understanding:\n\n| You want | Use | \n|---|---|\n| Answers to questions about your documents, with citations | PageIndex Agent | \n| The relevant pages, to feed your own LLM or agent | [Page Search API](https://pageindex.ai/search) | \n| Fields and tables as JSON, to store or process | [Page Extract API](https://pageindex.ai/extract) | \n\nPageIndex Agent also powers our own [Chat App](https://app.pageindex.ai/) and Page Extract.\n\n## \n\nPageIndex indexes and stores your documents; your own model searches the tree and writes the answer. That keeps you free to pick the balance of accuracy and cost that suits you, and to switch models as they improve, without re-indexing anything:\n\n``` python\nfrom pageindex import PageIndexClient\n\nclient = PageIndexClient(\n    index=\"cloud\",            # index and store in PageIndex Cloud\n    chat=\"gpt-6.1-terra\",     # your own model answers\n)\n\ndoc_id = client.submit_document(\"annual-report.pdf\", wait=True)[\"doc_id\"]\n\nfor chunk in client.chat(\"What drove revenue growth?\", doc_id=doc_id, stream=True):\n    print(chunk, end=\"\", flush=True)\n```\n\n`chat()` runs PageIndex's own document agent with your model. To use your own agent instead, take PageIndex's retrieval tools and system prompt into the framework you already use, in Python or JavaScript, or connect over MCP. See [Use with your LLM](https://pageindex.ai/docs/sdk/chat) and [Use with your agent](https://pageindex.ai/docs/sdk/agents).\n\n## \n\nInstall the SDK, create an API key, and ask your first question in a few minutes:\n\n```\npip install -U pageindex\n```\n\nOr paste this into Claude Code, Cursor, or any coding agent that can read a URL:\n\n```\nRead https://pageindex.ai/docs/SKILL.md and follow it to set up PageIndex in this project.\n```\n\nRead the [Quick start](https://pageindex.ai/docs/getting-started), see the [PageIndex Agent page](https://pageindex.ai/agent), or try it on real documents in our [examples](https://pageindex.ai/examples).", "url": "https://wpnews.pro/news/introducing-the-new-pageindex-agent", "canonical_source": "https://pageindex.ai/blog/pageindex-agent", "published_at": "2026-11-14 12:00:00+00:00", "updated_at": "2026-11-16 04:20:05.350070+00:00", "lang": "en", "topics": ["artificial-intelligence", "generative-ai", "ai-products", "ai-tools", "ai-search"], "entities": ["VectifyAI", "PageIndex Agent", "Page Search API", "Page Extract API", "Voyage 4 Large", "PageIndex"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/introducing-the-new-pageindex-agent", "markdown": "https://wpnews.pro/news/introducing-the-new-pageindex-agent.md", "text": "https://wpnews.pro/news/introducing-the-new-pageindex-agent.txt", "jsonld": "https://wpnews.pro/news/introducing-the-new-pageindex-agent.jsonld"}}