docling: Automated Document Parsing for Generative AI: From PDFs to Audio

Converts documents in various formats, including PDF, DOCX, PPTX, audio, and video files, into structured data. Going beyond simple text extraction, it precisely analyzes page layouts, table structures, formulas, and code blocks, organizing them into a format that is easy for generative AI to understand.

Using vision-language models like GraniteDocling and OCR, it can convert scanned documents and complex charts into tables or code. It supports integration with major agent frameworks such as LangChain and LlamaIndex, and runs in local environments, ensuring the secure processing of sensitive data.

Export URLs or local files to Markdown, HTML, JSON, and other formats with a single command. It serves as a highly useful standardized tool that reduces the complexity of the preprocessing stage when building RAG pipelines or developing document-based AI agents.