Sift AI: Modular Document Processing and Multi-Agent Debate Framework

A domain-first, AI-assisted research framework for structured document analysis and controlled multi-agent debate. Sift AI is a modular framework developed to explore how domain expertise, legal reasoning, and systems thinking can shape the behavior of large language models in complex analytical tasks. Originally developed to process legal texts, it has evolved into a general-purpose research environment for structured interpretation, argument testing, and debate simulation. The emphasis of Sift AI is not on the novelty of models, but on how they are framed, constrained, and coordinated.