CASE STUDY / 05
Make legal document analysis accessible, with a conversational memory that keeps a case's context.
A multi-agent architecture to analyze contracts, rulings and statutes — built at a hackathon.
- ROLE
- Multi-agent AI architecture
- SCOPE
- Legal RAG · Knowledge graph · Conversational memory
- STATUS
- HackLawDays 2025 hackathon
- YEAR
- 2025
PythonLangGraphRAGQdrantKnowledge Graph
01
CONTEXT
The problem
Analyzing a legal case — contracts, rulings, statutes — requires cross-referencing heterogeneous documents while keeping the thread of the case across multiple exchanges.
02
RESPONSE
The response
A multi-agent conversational architecture combining RAG with a legal knowledge graph, paired with a conversational memory system to keep a case's context across exchanges.
03
TRADE-OFFS
Design decisions
01
Dedicated knowledge graph
Model legal relationships instead of relying on vector search alone.
02
Conversational memory
Keep a case's context instead of treating every question in isolation.
04
STACK
Tech stack
PythonLangGraphRAGQdrantKnowledge Graph
NEXT PROJECT
ARIS