React Agent with Long-Term Memory and Tool Integration

This project demonstrates a React Agent built using LangGraph with advanced tool integration and long-term memory capabilities. The agent is designed to assist users in customer support scenarios with a rich set of tools and a sleek UI interface powered by Chainlit.

Features

• Knowledge Base Retriever: Retrieves information from a FAISS-based knowledge base.
• Customer Information Tool: Fetches user details for personalized support.
• Form Management:
  • Retrieve form details.
  • Assist users in completing forms step-by-step.
  • Submit forms and handle retries for any submission errors.
• Memory Capabilities:
  • Save Memory: Store contextual information during conversations.
  • Search Memory: Retrieve past interactions to ensure continuity and context.
  • Powered by Mem0AI for long-term memory management.
• Dynamic Chat UI: A user-friendly interface created with Chainlit for seamless interaction with the chatbot.

Project Setup

1. Install dependencies with Poetry:

   poetry install  

2. Setup environment variables: Provide required api keys: OPENAI_API_KEY=your-api-key.

   cp .env.example .env  

3. Start the local server:

   make start  

Access the API at http://localhost:8000

Architecture

The project utilizes LangGraph to define a conversational workflow, integrating tools and memory nodes. Below is an example diagram of the graph and the UI interface:

Graph Architecture

Chat UI

Sample conversations

1. Check order history of a customer:

   

2. Change user's address: