Feather-2/Burner-X

What it solves

Paper Burner X is an AI-powered workstation designed for researchers and deep learners to overcome the challenges of processing vast amounts of academic literature. It streamlines the workflow of reading, translating, and analyzing complex PDFs and other document formats while preserving formulas, tables, and original formatting.

How it works

The tool operates as a browser-based workstation that can be deployed as a pure frontend application (where data is stored locally in the browser) or via Docker. It integrates with various OCR engines (like MinerU and Doc2X) and AI model endpoints (such as DeepSeek, Gemini, and Claude) to perform high-speed concurrent translation and document analysis.

For long-text analysis, it implements a frontend-driven Agentic RAG system. This system provides the AI with a global view of the document's structure (a "layered meaning group/map") and a toolbox of tools like grep, vector search, and fetch to autonomously decide how to best extract information and reason across long documents.

Who it’s for

It is primarily aimed at graduate students, academic researchers, and knowledge workers who need to perform detailed, long-form reading and structured information extraction from academic papers.

Highlights

• Agentic RAG: A pure-frontend implementation of an AI agent that uses multi-step reasoning and tools to analyze long documents.
• High-Performance Translation: Supports concurrent processing of multiple files and high-speed translation of long papers in seconds.
• Format Preservation: Maintains original layout, formulas (LaTeX), and tables during translation and OCR.
• Multi-Format Support: Imports PDF, DOCX, PPTX, EPUB, Markdown, and content from GitHub or URLs.
• Privacy-First: Offers a pure-frontend mode where all data is stored locally in the browser (localStorage/IndexedDB).
• Academic-Specific Tools: Includes a "literature matrix" for structured data extraction and the ability to generate mind maps and Mermaid flowcharts.