CuroAI Library (GPT with long-term memory)

CuroAI is a JavaScript library designed for interfacing with the Curo AI system, which leverages OpenAI's powerful GPT-3.5 model for chat completions and Pinecone for efficient indexing and retrieval of chat history. This library simplifies the process of creating conversational agents that can remember and respond to previous interactions intelligently.

This library tries to mimic a memory between the system and user beyond a conversation, or simply put a pseudo-memory.

Installation

To use CuroAI in your JavaScript project, you can install it using npm or yarn:

npm install curoai
# or
yarn add curoai

Initialization

1. Import the library into your project:

import CuroAI from 'curoai';

2. Create an instance of the CuroAI client by providing your OpenAI API Key and Pinecone API Key:

   const curoai = new CuroAI({
    openAIAPIKey: 'YOUR_OPENAI_API_KEY',
    pinecone: {
    environment: 'YOUR_PINECONE_ENVIRONMENT',
    apiKey: 'YOUR_PINECONE_API_KEY'
    },
   // Optionally, specify the initial OpenAI model (default: 'gpt-3.5-turbo-16k')
    model: 'gpt-3.5-turbo-16k',
   });

Usage

Sending messages

You can use CuroAI to send messages in a conversation. Messages should be provided as an array of objects with a role and optional content:

const messages = [
  { role: 'user', content: 'Hello, CuroAI!' },
  { role: 'assistant', content: 'Hi there! How can I assist you?' },
];

const user = {
  uid: 'USER_ID', // Unique reference to the user interacting with the system
  firstName: 'John', // User's first name
};

const input = {
  messages,
  user,
};

// Get a response as a stream of chat chunks
const response = await curo.getCompletionStream(input);

// The reponse by default is a stream
console.log(response);

Notes

• The CuroAI library facilitates intelligent conversations by remembering and referencing previous interactions.
• Pinecone is used to efficiently index and retrieve chat history.
• Remember to replace 'YOUR_OPENAI_API_KEY', 'YOUR_PINECONE_ENVIRONMENT', and 'YOUR_PINECONE_API_KEY' with your actual API keys.
• You can customize the OpenAI model used by modifying the model property during client initialization.

And, feel free to pitch in ideas on how to improve this!