A better legal module inside fudge would preserve provenance, connect arguments to sources, and let professionals inspect how a conclusion was assembled.
Why this matters
Document-heavy workflows bury signal under versioning, duplication, and poor recall.
Generic AI tools sound authoritative even when they cannot show the chain of reasoning behind a conclusion.
Privacy, client sensitivity, and auditability make pure cloud dependency a poor fit for many legal settings.
How we approach it
fudge provides the local-first document and workspace layer for sensitive legal work.
samar adds a model for claims, evidence, contradiction, and confidence that is more rigorous than generic retrieval-plus-chat.
Future Wiyc data infrastructure can improve legal-specific retrieval, review, and drafting workflows without pretending to replace legal judgment.
Where things stand
No legal module exists today.
This page should be read as a thesis for where fudge can go once its foundations are stronger.
Branches in play
Software. Supports controlled document environments and long-lived case context.
Data. Improves the quality of legal corpora and task-specific training signals.
Intelligence. Tracks source confidence, conflict, and reasoning paths.
People. Ensures interpretation remains accountable to human experts.
The legal opportunity is not another confident assistant. It is a calmer, more exact domain layer inside fudge.