Document Classification With Natural Language Rules

via “document-classification-with-natural-language-rules”

A data framework for building LLM applications over external data.

Unique: Uses natural-language rule definitions interpreted by LLMs instead of code-based rules or machine learning models, enabling non-technical users to define and modify classification logic without programming, while supporting semantic understanding of document content

vs others: More flexible than rule-based systems (no regex required) and more interpretable than machine learning classifiers (rules are human-readable), but slower and more expensive than both due to per-document LLM inference