anishasingh-dev/openenv-ai-support-simulation-projects

🧠 OpenEnv: AI Customer Support Simulation

📌 Overview

OpenEnv is a simulation environment for building and evaluating AI agents in a customer support setting.

The goal is to design an agent that can intelligently process incoming support tickets by:

• —🏷️ Classifying the issue category
• —⚡ Determining the priority level
• —🎯 Selecting the appropriate support action

The environment evaluates the agent using a deterministic reward function, making it ideal for experimentation, benchmarking, and learning.

Tech Stack

• —python
• —AI agent logic
• —Simulation environment

📂 Dataset

The dataset consists of simulated customer support tickets that mimic real-world scenarios, including:

• —💳 Billing issues
• —🛠️ Technical problems
• —❓ General queries
• —🤯 Messy / real-world user inputs (typos, urgency, informal tone)

Each ticket includes ground truth labels for:

• —Category
• —Priority
• —Action

🧮 Reward Function

The agent is evaluated per ticket using the following scoring:

✅ Maximum score per ticket = 1.0

▶️ How to Run

Run Baseline Agent

python baseline_agent.py

🐳 Docker Setup

Build the Container

docker build -t support-env .

Run the Container

docker run support-env

🗂️ Project Structure

.
├── dataset.py          # Dataset definitions
├── env.py              # Simulation environment
├── grader.py           # Evaluation logic
├── baseline_agent.py   # Simple rule-based agent
├── requirements.txt    # Dependencies
├── Dockerfile          # Container setup
└── README.md           # Project documentation

📊 Example Output

Final Score: 20.10

🚀 Future Improvements

• —Add ML/NLP-based intelligent agents
• —Introduce noisy / adversarial inputs
• —Expand dataset with more edge cases
• —Add web-based visualization dashboard

🤝 Contributing

Contributions are welcome! Feel free to fork the repo, open issues, or submit pull requests.