Interactive Planning for Orchestrated Multi-Agent Systems

AIPOM is a system that integrates natural language interaction with direct manipulation of plan graphs to support human-in-the-loop planning in orchestrated multi-agent systems (OMAS). The system enables users to transparently inspect, refine, and collaboratively guide LLM-generated plans, enhancing user control and trust in multi-agent workflows.

This code is accompanying a paper submission to the ACL 2025 System Demonstration. It represents a limited version of the original system, tailored for math reasoning problems, with a subset of agents due to external API access issues. Please refer to the paper for the full system details and implementation.

🎬 Demo Video

Click below 🖼️ to see AIPOM in action:

Prerequisites

Before installing, make sure you have the following installed:

• Python (version 3.8 or higher)
• Node.js (version 16 or higher)
• npm (version 7 or higher)

OpenAI API Key

AIPOM is powered by GPT, and requires an OpenAI API key and organization to use the system. You need to set these values before running the project.

1. In your terminal, run the following commands to set your OpenAI API key and organization ID:

   export OPENAI_API_KEY="your-api-key-here"
   export OPENAI_ORGANIZATION="your-organization-id-here"

2. Verify that the environment variables are set correctly by running:

   echo $OPENAI_API_KEY
   echo $OPENAI_ORGANIZATION

Installation

Backend (Python)

1. Clone the repository:

   git clone https://github.com/rit-git/aipom.git
   cd aipom

2. Create conda env

   conda create -n <env_name> python=3.12.3

3. Install the required Python dependencies:

   pip install -r requirements.txt

Frontend (React)

1. Navigate to the frontend directory:

   cd frontend

2. Install the required Node.js dependencies:

   npm install

Usage

To start the development environment, run both the Python backend and React frontend in separate terminals:

Terminal 1: Start the Python backend server

python server.py

This starts the backend server, which handles the logic for plan generation, refinement, and agent coordination.

In case of ERROR: [Errno 48] Address already in use while running server

> sudo lsof -i:8000
> kill -9 PID

Terminal 2: Start the React frontend server

cd frontend
npm run dev

This starts the frontend server. Open your browser and go to http://localhost:5173.

If you encounter any issues or the system is not responding, try refreshing your browser.

Guidelines

We recommend following the guidelines below, for testing our system prototype:

1. Restrict query to math reasoning questions
2. While configuring agent params, restrict model to one of [gpt-4o, gpt-4o-mini]