How mekalakarthik05/lexintel-ai works
LexIntel AI is a legal-document research assistant: users upload documents for chunking and indexing, then ask natural-language questions and receive answers grounded in retrieved document context. The Streamlit interface calls a FastAPI backend, which manages sessions and routes uploads and chat requests into the RAG components. The README documents the embedding, Pinecone retrieval, context-injection, and Llama/Groq stages; some internal wiring is not fully shown in the excerpts, so those documented links cite the README.
Main entry points: vector_store.py Vector Store · src/app.py FastAPI Routes · chatbot.py Legal Assistant · app.py Streamlit Interface
Components 13
User Interface
- Streamlit Interface
app.py
API and Sessions
- FastAPI Routes
src/app.py - Session Manager
src/session_manager.py
Document Ingestion
- Document Processor
src/document_processor.py - Text Chunking
src/document_processor.py - Document Uploader
document_uploader.py
Retrieval and Answering
- Embedding Model
src/vector_store.py - Vector Store
src/vector_store.py - Pinecone Index
- Legal Assistant
src/chatbot.py - Runtime Config
src/config.py
Other components
- Researcher
- Groq LLM
Connections 14
- Researcher → Streamlit Interface: asks or uploads
README.md - Streamlit Interface → FastAPI Routes: sends requests
app.py - FastAPI Routes → Document Processor: processes uploads
src/app.py - Document Processor → Text Chunking: splits documents
src/document_processor.py - FastAPI Routes → Vector Store: indexes documents
README.md - Document Uploader → Vector Store: uploads vectors
- FastAPI Routes → Session Manager: manages sessions
src/app.py - FastAPI Routes → Legal Assistant: routes questions
src/app.py - Legal Assistant → Vector Store: retrieves context
README.md - Vector Store → Embedding Model: embeds queries
README.md - Vector Store → Pinecone Index: searches vectors
README.md - Legal Assistant → Groq LLM: generates answers
README.md - Legal Assistant → Runtime Config: reads credentials
src/chatbot.py - Vector Store → Runtime Config: reads index settings
src/vector_store.py