Jev.aitools.fyi

Command Palette

Search for a command to run...

DemoFeatured
Product profile
PythonUpdated 2026-09-20

DocJev

OSS document classify and split library powered by Jev and natural-language category rules, with liteparse or LlamaParse backends.

Overview

DocJev is an open-source library from Jerry Liu (github.com/jerryjliu/docjev) for document classification and splitting with TypeSafe Jev. Supply natural-language category rules alongside document text; Jev returns a predicted category or sub-document boundaries.

Problem: Document pipelines need fast category labels and split boundaries without paying chat-model prices on every page.

Built for: Teams building ingestion, RAG, or compliance flows over PDFs and scans who want rule-driven classify and split steps.

First indexed on Jev Directory: 2026-09-20

Creator and team

Name
Jerry Liu
Organization
LlamaIndex
Handle
@jerryjliu0
“Give a document alongside natural language category rules. Jev will predict the document category or the boundaries between sub-documents.”

How Jev is used

Role in the product flow
Document classification and split-boundary prediction
Primitives
ChoiceScore
State in
Parsed document text from liteparse or LlamaParse plus user-authored natural-language category or split rules from the DocJev API.
Decision out
Predicted document category for classify mode, or boundary decisions between sub-documents for split mode.
  1. Parse document with liteparse (OSS) or LlamaParse (VLM preprocessing)
  2. Format rules and text for Jev System One calls
  3. Jev classifies or marks split boundaries per DocJev configuration

DocJev wraps Jev as the decision core for both classify and split workflows so teams express policy in natural language instead of brittle regex. liteparse is the default fast OSS parser for digital documents; LlamaParse adds VLM preprocessing latency but suits cached digitization for downstream LlamaIndex tasks. Confirm latest benchmarks and APIs in the DocJev repository.

Sourced performance claims

  • About six times faster than gpt-5.6-luna with equivalent accuracy in the launch benchmark (liteparse timings included).Source: Jerry Liu DocJev announcement on X

Features and stack

Features

  • Open source at github.com/jerryjliu/docjev
  • Classify and split modes with natural-language rules
  • liteparse integration (github.com/run-llama/liteparse)
  • Optional LlamaParse via cloud.llamaindex.ai

Stack

  • DocJev
  • TypeSafe System One
  • liteparse
  • LlamaParse
  • LlamaIndex ecosystem

Demo

Showcase clip with the same lightbox player as the homepage. Click to play.

Jerry Liu@jerryjliu0

DocJev classify and split

Feed a doc plus plain-English category rules. DocJev uses Jev to classify or find split boundaries six times faster than gpt-5.6-luna at matching accuracy, liteparse included.

productdocumentsclassification

FAQ

Is DocJev part of LlamaIndex core?
DocJev is a standalone OSS repo from Jerry Liu. It integrates optional LlamaParse and liteparse from the Run Llama ecosystem; treat github.com/jerryjliu/docjev as the source of truth.

Related learn guides

Original Jev guidance that pairs with this product pattern.

  • Jev use cases

    The patterns builders actually search for: moderation, routing, triage, RAG verify, and agent gates.

  • Jev vs LLM classification

    When to gate with System One probabilities instead of asking a chat model to label things.

Related products

Hand-picked neighbors with rich profiles or overlapping tags.

IntegrationsPython2
LlamaIndex reranker + router powered by TypeSafe Jev, typed scores/choices, cheaper than LLM-as-judge.
integrationsroutingrag
ApplicationsTypeScript
FeaturedDemoMCP
classifier.
applicationsclassificationmoderation
ApplicationsJavaScript
Demo
Check whether each cited paper supports the sentence citing it.
applicationsagents
By Rishit