whereismyai
Jev Doc Search
Jev Doc Search is a search tool that finds the exact page answering a question in long documents using Jev choices and PageIndex.
Jev Doc Search finds the page that answers a question in long reports and documents. It avoids vector databases and embeddings by using Jev choices to evaluate multiple-choice options and PageIndex to represent documents as hierarchical trees. The software builds a tree from the document structure containing sections, summaries, and pages. Jev searches the tree step by step, picking the most likely child nodes until it reaches the target pages. It supports deeper trees, top-K search with beam search, and checking candidates with Noul yes/no questions. It runs via Python scripts like pagesearch.py and treesearch.py, requiring API keys from the TypeSafe console and PageIndex dashboard.
Jev Doc Search at a glance
- Best for: Software engineer, Researcher
- Runs on: API, self-hosted, open source
- Pricing: Open source under the Apache 2.0 license
- Released: September 2026
- Category: AI search engine
- Website: github.com/vectifyai/jev-doc-search
Key features
- Finds answer pages in long documents
- Builds hierarchical document trees
- Evaluates choices with probabilities
- Performs top-K beam search
- Checks candidates with Noul yes/no questions
- Uploads PDFs and reuses document IDs
Use it for
- Locate revenue figures in corporate annual reports
- Search through employee handbooks
- Find net income data in financial filings
Alternatives to Jev Doc Search
- Jev Bisect — It uses exact rational arithmetic and integer candidate indexes for numeric searches.
- XERJ — It answers agent questions from an index instead of wasting tokens on context windows.
- Hister — Protects privacy with a self-hosted search index for your pages and files.
- Jina — It provides reliable reader and search APIs optimized for LLMs and RAG systems.
- Exa — Provides high-accuracy web search and deep research endpoints for AI agents.
- jurl — Extracts exact page content without rewriting or hallucinating answers.
- PixelRAG — Preserves visual structure like tables and charts that text parsers miss.
- Searchbase — Independent web access with support for REST, MCP, self-hosting, and managed cloud.
- Parallel — It gives AI agents real-time access to fresh, cited web context through a single API.
- Crawl4AI — Turns any URL into clean Markdown or typed JSON for AI models.
- MemFree — It combines hybrid web search with AI summarization and page generation.
- Shlexball — It runs locally on your device to keep files private while searching content.
All Jev Doc Search alternatives