GraphRAG Chat

An advanced chat application for interacting with call transcripts using a scalable and efficient graph-based RAG approach.

GraphRAG Chat

  • AI Chatbot

  • RAG

  • Call Transcripts

  • Neo4j

About the Project

GraphRAG Chat is a cutting-edge chat application developed in my role as a Generative AI and Backend Developer at Fixit AI. This system enables users to interact with call transcripts efficiently using a Retrieval-Augmented Generation (RAG) approach. Leveraging the power of graph-based knowledge retrieval, it provides highly contextual responses while maintaining memory across interactions. Designed for scalability, GraphRAG Chat is optimized for enterprise-level use cases.

Key Features

  • Graph-Based RAG: Utilizes Neo4j to structure and retrieve knowledge efficiently.
  • Context-Aware Conversations: Retains memory to provide relevant responses over time.
  • Scalability & Efficiency: Optimized for handling large volumes of transcripts.
  • Multi-Turn Query Handling: Understands complex queries across multiple interactions.
  • Real-Time Insights: Extracts key insights from call transcripts for quick decision-making.

Screenshots

Screenshot 1
A user interacting with call transcripts in a conversational format.

Screenshot 2
Graph visualization of connected call data in Neo4j.

Technologies Used

  • LLM Framework: LangChain
  • Database: Neo4j (Graph Database)
  • Memory Layer: Persistent memory for maintaining context
  • Backend: FastAPI
  • Storage: Cloud-based transcript storage for scalability
  • Authentication: OAuth & JWT for secure access

Developed as part of my work at Fixit AI, GraphRAG Chat revolutionizes how users interact with call transcripts. 🚀