LlamaIndex explains just-in-time Agentic OCR: how two-pass document processing balances cost and accuracy
Original title:Just-in-Time Agentic OCR
AI summary
The LlamaIndex blog proposes a just-in-time Agentic OCR pattern: first use a free parser such as LiteParse to roughly read all files for retrieval, then call a VLM to perform OCR only on relevant pages.
Quick Facts
- Published
- Updated
- Source
- LlamaIndex:产品、工程与评测
- Category
- Tutorials & Tips
- AI score
- 65
作者用 84 份 SEC 文件实测两遍式文档处理流程,给出明确的适用边界和成本数字,方法可直接迁移到自己的数据室场景。
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: LlamaIndex explains just-in-time Agentic OCR: how two-pass document processing balances cost and accuracy. September 11, 2026. https://new.nat6.net/en/news/cmtym2c3q0003rob0w6kxvnn3/