OpenAI Conversation Builder

This project is a Node.js script that simulates a conversation between two agents using OpenAI's API. It uses gpt-4o-mini to generate an iterative conversation, including follow-up questions and responses.

Features

• Automatically generates a conversation between two agents.
• Logs conversation progress to a file in the conversations directory.
• Tracks token usage and estimates costs for both input and output tokens.
• Supports dynamic follow-up question generation based on the current conversation context.
• Differentiates between cached and non-cached input tokens for accurate cost estimation.
• Outputs system instructions and prompts for better transparency of conversation flow.

Requirements

• Node.js
• The following npm packages:
  • axios
  • dotenv
  • fs
  • chalk

Setup Instructions

1. Clone the repository and navigate to the project directory.

2. Install dependencies by running:

   npm install axios dotenv chalk

3. Create a .env file in the root of the project directory and add your OpenAI API key:

   OPENAI_API_KEY=your_openai_api_key_here

4. Run the script using the following command:

   node conversation_builder.js

Script Overview

• The script starts a conversation with an initial question, such as:

  How will AI shape the future of the world?
  

• It then generates responses and follow-up questions iteratively by making calls to OpenAI's API.

• After each response, the script saves the updated conversation to a JSON file in the conversations directory.

• System instructions for the answering agent and the follow-up prompts for the questioning agent are displayed at the start of the conversation for better transparency.

• The script estimates the costs for input, cached input, and output tokens based on the following rates:

  • Input Tokens: $0.15 per million tokens
  • Cached Input Tokens: $0.075 per million tokens
  • Output Tokens: $0.60 per million tokens

• After each API call, the total input, cached input, and output tokens are tracked, and the estimated cost is displayed.

Cost Estimation

• Input Tokens: $0.15 per million tokens
• Cached Input Tokens: $0.075 per million tokens
• Output Tokens: $0.60 per million tokens

The script will provide real-time cost estimation based on the number of tokens used for each API call, including any cached tokens that reduce the overall cost.

Folder Structure

• conversations/ - Directory where the conversation files are saved.

Notes

• The script will create the conversations directory if it does not exist.
• The script runs for a specified number of turns (numberOfMessages), simulating a conversation between the curious agent and the answering agent.
• The script supports up to 10 conversation turns by default, but this can be adjusted in the code by modifying the numberOfMessages variable.
• The final follow-up question is not sent, as there will be no response generated for it.

License

This project is licensed under the MIT License.