Skip to content
FUNFACTOR1Public

About

High-precision Graph-RAG legal engine for Italian Law. Features dynamic synchronization, deterministic validity checks, and intelligent cross-code navigation to eliminate LLM hallucinations.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

3 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 

Repository files navigation

Juris AI is a next-generation legal assistant built to bridge the structural gaps found in general-purpose LLMs and standard RAG (Retrieval-Augmented Generation) setups. While traditional systems often rely on "frozen" or fragmented knowledge, Juris AI uses a Hybrid Graph-RAG architecture to ensure every response is anchored exclusively to current, verified, and in-force legislation. 🚀 Vision & Problem Solving

In the legal field, "plausible" isn't good enough—it's a liability. Standard LLMs typically struggle with three critical pain points:

Legislative Stratification: Difficulty managing articles with complex suffixes (e.g., bis, ter, quinquies) or identifying repealed sections.

Lack of Systemic Overview: An inability to logically connect cases across different codes (such as linking Criminal proceedings to Civil liability).

Data Stagnation: A reliance on static databases that fail to account for last-minute regulatory updates.

Juris AI solves these challenges by decoupling the presentation layer from a deterministic source validation engine. ✨ Key Capabilities

Dynamic Knowledge Sync: Unlike static tools, our engine is designed to realign the entire knowledge base with the latest legislative state at the start of every session.

Real-Time Validity Checks: Every retrieved article undergoes a metadata analysis to confirm its status (effective vs. repealed) before it ever reaches the user.

Intelligent Cross-Code Navigation: The system maps the intricate dependencies between criminal penalties and civil compensation, ensuring a complete and structured reconstruction of the remedial framework.

Full Authoritative Text Retrieval: We reject "creative paraphrasing." Juris AI provides the literal text of the law, ensuring legal opinions are grounded in the actual letter of the statutes.

🛠 High-Level Architecture

The engine runs on a proprietary orchestration layer that coordinates three core pillars:

Knowledge Graph (IFLA LRMoo): A relational map modeling hierarchies and cross-references between norms (using a Work/Expression/Manifestation logic).

High-Precision Vector Store: Powered by BAAI/bge-m3 embeddings for advanced multilingual semantic search.

Deterministic Orchestrator: The "brain" of the system that enforces strict logical constraints on data retrieval, effectively eliminating hallucinations and ensuring every citation is structurally verified.

📊 Benchmarks & Results

We put Juris AI through a rigorous stress test against leading generalist models:

Hallucination Rate: 0% on tested repealed norms (e.g., Art. 155-quinquies C.C., which standard models often mistake as still being in force).

Multi-hop Reasoning: Perfect reconstruction of remedial sequences (e.g., connecting Art. 2043 to Arts. 2056-2059 C.C.).

Surgical Precision: Instant identification of articles with suffixes and recent legislative novellas.

👉 Download the Full Benchmark Report (PDF) 👉 Watch the Demo Videos on the Official Site 📞 Get in Touch

Juris AI is an independent project with a proprietary architecture. We are currently open to discussions regarding:

Technical collaborations on Graph-RAG architectures.

Professional licensing for law firms or compliance departments.

Adapting the engine for other national legal systems.

Developer: Zampier Zago

Version: 3.0.0

About

High-precision Graph-RAG legal engine for Italian Law. Features dynamic synchronization, deterministic validity checks, and intelligent cross-code navigation to eliminate LLM hallucinations.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors