LangMem Customer Support Agent

A customer support agent with memory that automatically classifies inquiries, creates tickets, and maintains conversation history using LangGraph and OpenAI.

Quick Start

Install

git clone <repository-url>
cd langmem-customer-support-agent
pip install -r requirements.txt
cp env.example .env
# Add your OpenAI API key to .env

Run

python main.py

Usage

Choose from the menu:

1. Custom ticket - Enter your own support request
2. Example ticket - Test with predefined scenarios
3. Search memory - Find previous customer interactions
4. Exit

Example

Step 1. Choose option: 2 (Example ticket)
Step 2. Choose example: 1 (Login issues)

✅ Agent automatically:
- Searches for customer history
- Creates support ticket
- Sends personalized reply
- Stores interaction in memory

Choose option: 3 (Search memory)
Enter query: login issues

✅ Found previous interaction:
- Customer: test@test.com
- Issue: Login problems
- Status: Awaiting response

How It Works

1. Triage Agent classifies incoming messages (ignore/notify/respond)
2. Response Agent processes requests using tools (create tickets, send replies)
3. Memory System automatically stores all interactions for future reference
4. Search finds relevant customer history using semantic similarity

Agent Components

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│  Triage Agent   │───▶│ Response Agent  │───▶│  Memory Store   │
│                 │    │                 │    │                 │
│ • Classifies    │    │ • Uses Tools    │    │ • InMemoryStore │
│ • Routes        │    │ • Sends Replies │    │ • Vector Search │
│ • Updates State │    │ • Creates Memory│    │ • Persistence   │
└─────────────────┘    └─────────────────┘    └─────────────────┘

Project Structure

├── agents/          # Agent workflow and classification
├── tools/           # Memory, ticket, and reply tools
├── prompts/         # System prompts and instructions
├── tests/           # Test scenarios
└── main.py          # Interactive interface

Requirements

• Python 3.11+
• OpenAI API key
• LangGraph 0.4.0+

---

Built with LangGraph, LangChain, and OpenAI