tznurmin / robofriend

RoboFriend is an AI-powered penpal application using OpenAI's GPT and Docker. Engage in meaningful email exchanges, automatically receive personalized responses, and enjoy seamless email management with a persistent MongoDB-backed conversation history.

RoboFriend

RoboFriend is a Dockerized AI-powered penpal application. It uses an AI model to generate engaging email responses and a mailer service to send and receive emails. The application is composed of three main services:

1. AI Service: Uses an AI model to generate email responses and summarize discussions.
2. Mailer Service: Sends and receives emails, managing the communication with your penpal.
3. MongoDB service: Stores the email exchanges and current summary between RoboFriend and the user.

Project Structure

Here's an overview of the project structure:

docker
├── df
│   ├── Dockerfile.ai              # Dockerfile for the AI service
│   ├── Dockerfile.mailer          # Dockerfile for the mailer service
│   ├── ai_data
│   │   ├── ai_penpal.py           # AI penpal logic
│   │   ├── discussion_summary.py  # Discussion summarization logic
│   │   ├── maildb.py              # Database interaction module for AI service
│   │   ├── openai.key             # API key for OpenAI GPT model
│   │   └── requirements.txt       # Python dependencies for the AI service
│   └── mailer_data
│       ├── maildb.py              # Database interaction module for mailer service
│       ├── penpal_mailer.py       # Email sending and receiving logic
│       ├── requirements.txt       # Python dependencies for the mailer service
│       └── token.json             # OAuth2 token for Google API
├── docker-compose.yaml            # Docker Compose configuration file
└── init-mongo.sh                  # Initialization script for MongoDB

Requirements

• Docker
• Docker Compose
• A Google API OAuth2 token for accessing Gmail (follow these instructions to obtain a token)
• An OpenAI API key for using the GPT model (sign up on OpenAI's platform to get an API key)

Setup

1. Clone this repository:

   git clone https://github.com/tznurmin/robofriend.git
   cd robofriend/docker
   

2. Add your OpenAI API key to df/ai_data/openai.key.

3. Add your Google API OAuth2 token to df/mailer_data/token.json.

4. Edit the docker-compose.yaml file to configure the project details.

5. Build and run the application using Docker Compose:

   docker-compose up -d
   

6. Your RoboFriend is now up and running! The AI service will start generating email responses and the mailer service will send and receive emails.

Usage

Once the application is running, simply send an email to the configured email address. In the email address's local part (the part before the '@' symbol), include a plus sign (+) followed by a user ID to uniquely identify the user. This helps the AI service differentiate between different users and maintain separate conversation histories for each user. The AI service will generate a response based on the summary stored in the MongoDB service, and the mailer service will send the email to the recipient.

For example, if your configured email address is [e-mail] and the user ID is john123, the email should be sent to [e-mail].

Stopping the Application

To stop the running containers and remove the associated resources, execute the following command in the project directory:

docker-compose down

License

This project is licensed under the MIT License. See the LICENSE file for more details.