ai-team

AI Team with Memory

AI Team is a group of expert personas you can interact with:

James Ryan Wick: The ultimate software development expert—a perfectionist, innovative, and endlessly passionate about creating human-readable, flawless code.
Jimmy Roar: A hilariously quirky motivational trainer with boundless energy who can lift anyone’s spirits with humor and insight.
Neil Knows: The ultimate marketing guru—insightful, strategic, and creative, with unmatched knowledge of sales, advertising, and SEO.
Sophia Akasha Lightray: An energetic healer and spiritual guide, blending deep wisdom, compassion, and expertise in holistic practices.
Dr. Sigmund Leud: A psychologist with a peculiar and eccentric personality, inspired by Sigmund Freud, who loves to interpret dreams and explore the mind.

Note: You can also trigger communication between the AI characters for collaborative responses.

Example Outputs

How It Works

The AI characters are built using llama3.2:3b and qwen2.5-coder models as the foundation, but you can customize them as needed. The system is powered by Ollama and supports:

Individual session histories to maintain context during conversations.
A persistent database to store chat histories and retrieve information using embeddings, enabling the models to remember details like your name or preferences.

Language Support: Chats in German are supported, but the responses may not always be perfect. For optimal results, using English is recommended.

Model Recommendations: While qwen2.5-coder provides good suggestions, larger models are recommended for handling more complex tasks or providing more detailed insights.

Preparation

Install Ollama

Download and install Ollama from their official website.

Create a Virtual Environment

In the root directory of the project:

python -m venv .venv
source .venv/bin/activate  # For Linux/macOS
.venv\Scripts\activate   # For Windows

Install Dependencies

Run the following command to install the required Python packages:

pip install -r requirements.txt

Install Models and Embeddings

Install via Script

Run the following command to download the models and create the specific configurations:

./create_models.sh

Usage

Start the Ollama Server

Ensure the Ollama server is running. Start it with:

ollama serve

Run the Script

Start the app using:

python app.py

Open main.html in your browser.

Interaction:

Click on an AI character to chat with it. Type in the Ask textbox and hit Send.
Trigger AI-to-AI communication using CTRL + click on the AI character you want to receive the trigger message. Then type your input and hit Send. To stop AI-to-AI communication, click the Stop button.

Additional Information

Install Models Manually

Run the following commands to download the models:

ollama pull llama3.2:3b
ollama pull qwen2.5-coder:7b
ollama pull nomic-embed-text

Run these commands to create the AI characters using the provided model files:

ollama create james_ryan_wick -f modelfiles/Modelfile-james_ryan_wick
ollama create jimmy_roar -f modelfiles/Modelfile-jimmy_roar
ollama create neil_knows -f modelfiles/Modelfile-neil_knows
ollama create sophia_akasha_lightray -f modelfiles/Modelfile-sophia_akasha_lightray
ollama create dr_sigmund_leud -f modelfiles/Modelfile-dr_sigmund_leud
ollama create zoe_voss -f modelfiles/Modelfile-zoe_voss

Offline Libraries

Note: To ensure offline functionality, download all necessary libraries and resources (e.g., jQuery, Bootstrap, Highlight.js) and replace the URLs in main.html with local paths.

Legal Information

This tool uses open-source models and embeddings. Refer to their respective licenses for compliance.
Ensure your use case complies with data protection regulations, particularly when handling sensitive or personal information.
The authors provide no warranty and assume no liability for any issues arising from the use of this tool.