Document Parsing Overview

Updated at:

Learn how to use Parse to convert complex documents into structured results that are easy to retrieve, display, and process downstream.

Document parsing organizes content from text-and-image files, audio, and video into easy-to-read results, providing Agents with the content they need to understand materials.

What Problems Does Parse Solve

Materials you haveWhat parsing providesTypical uses
Text-and-image materials such as PDFs, Office documents, and imagesBody text, headings, tables, image descriptions, and moreOrganize materials, prepare knowledge base content, prepare inputs for subsequent field extraction
AudioSpeech content with corresponding time positionsReview recordings and locate passages to replay
VideoVisual- and speech-related content, timeline, story segments, and summariesBrowse training recordings, understand video content, and locate segments to replay

For example, with a training document, you can parse the full text first and then check whether headings, tables, and figures are complete. With a training video, you can read the summary to grasp the topic and then use the timeline to replay the key parts. Different processing settings determine which auxiliary information the results contain; see Configure Document Parsing for details. For file formats and input conditions, see Supported Files and Limits.

Get Started

  • Use the Playground Console: After obtaining access, enter through the officially authorized entry and learn about inputs, outputs, and enhancement options in the console.
  • Configure Document Parsing: Configure file information, parsing scope, enhancement options, and result formats.
  • Get Document Parsing Results: Poll and validate results, and build stable storage and quality-check processes.
  • Use the CLI: Submit parsing tasks and query results in the terminal with bl parsex parse.