Deterministic Text Decomposer

Example Use Cases

• 1

  Preprocess and classify complex documents into structured semantic units without LLM inference costs.

• 2

  Filter and route text based on risk and authority to optimize downstream LLM processing.

• 3

  That require fast, deterministic, and offline text decomposition to improve document understanding and reduce token usage.

Description

Decompose is a fast, offline, and deterministic text classification tool that breaks down any text into classified, structured semantic units using regex and heuristics. It enables AI agents to preprocess documents by extracting standards, scoring risks, and filtering important content before passing it to LLMs, reducing token usage and improving reasoning efficiency. Built by Echology and originally part of AECai, it supports various domains and offers integration as a CLI, library, or MCP server.