AI Chatroom

A multi-agent AI discussion platform where various AI models can engage in conversations about specified topics. This project allows you to observe AI thinking processes and gather insightful answers from multi-agent discussions.

Features

• Create discussion rooms with customizable topics
• Configure multiple AI agents with different roles and personalities
• Support for OpenAI and Gemini models
• Web search capabilities for agents to find real-time information
• Real-time discussion visualization
• Configurable discussion length

Technology Stack

• Framework: Camel AI for multi-agent conversations
• Backend: FastAPI
• Frontend: HTML, JavaScript, Bootstrap
• AI Models: Support for OpenAI GPT models and Gemini models
• Search Tools: Integration with Tavily, Exa, and DuckDuckGo for web search

Installation

1. Clone the repository:

git clone <repository-url>
cd AI-chatroom

2. Install the required dependencies:

pip install -r requirements.txt

3. Set up your API keys:

Copy env.example to .env and fill in your API keys:

cp env.example .env
# Edit .env with your API keys

The application uses the following API keys:

• OPENAI_API_KEY: Required for OpenAI models (GPT-4o, GPT-4o-mini)
• GEMINI_API_KEY: Required for Google models (Gemini 2.5 Flash Preview, Gemini 1.5 Pro)
• TAVILY_API_KEY: For Tavily search (highest priority, optional)
• EXA_API_KEY: For Exa search (second priority, optional)

If no search API keys are provided, DuckDuckGo will be used as a fallback (no API key required).

Supported Models

The application currently supports the following AI models:

OpenAI Models

• GPT-4o
• GPT-4o Mini

Gemini Models

• Gemini 2.5 Flash Preview
• Gemini 1.5 Pro

If you attempt to use a model without providing the corresponding API key, the system will automatically fall back to GPT-4o-mini.

Usage

1. Start the application:

python main.py

2. Open your browser and navigate to http://localhost:8000

3. Create a new chatroom:

   • Enter a topic title
   • Add AI agents with names, models, and roles
   • Enable search capabilities for agents that need real-time information
   • Set the maximum number of discussion turns
   • Click "Create Discussion"

4. In the chatroom view:

   • Click "Start Discussion" to begin the AI conversation
   • Watch as the AI agents discuss the topic
   • Agents with search capabilities will use web search when needed
   • Search usage is indicated with a search icon next to the message
   • The discussion will automatically progress through the specified number of turns

Search Tool Configuration

The application prioritizes search tools in the following order:

1. Tavily Search: Used first when TAVILY_API_KEY is provided
2. Exa Search: Used when EXA_API_KEY is provided
3. DuckDuckGo: Used as a fallback when no other search APIs are available

If a primary search tool fails (e.g., due to rate limiting or API issues), the system automatically falls back to DuckDuckGo.

Search tool usage is logged in the logs/search_logs.json file for monitoring and debugging purposes.

Extending the Project

Adding New AI Models

To add support for additional AI models:

1. Update the create_model_from_string function in models.py to include the new model in the mapping
2. Add the necessary API key to your .env file
3. Update the frontend model dropdown in templates/index.html to include the new model option

For example, to add a new model called "new-model" from a provider called "new-provider":

# In models.py
model_mapping = {
    # Existing models...
    
    # New model
    "new-model": (ModelPlatformType.NEW_PROVIDER, ModelType.NEW_MODEL, NewProviderConfig().as_dict()),
}

Adding New Search Tools

To add support for additional search tools:

1. Update the create_search_tools function in models.py
2. Add the necessary API keys to your .env file
3. Modify the prioritization logic as needed

Customizing the UI

The UI templates are located in the templates directory and can be modified to change the appearance and functionality of the application.

License

MIT