Maestro-Orchestrator: Pluralising Synthetic Intelligence

Maestro-Orchestrator is a state-of-the-art orchestration engine designed to unify a diverse range of AI agents through a structured system of synthetic consensus and dissent. This innovative platform allows for real-time prompt routing among a rotating council of large language models, each selected for their unique capabilities. The output of these models is synthesized using quorum logic, ensuring balanced and effective decision-making.

🚀 Key Features

• FastAPI Backend: Provides live orchestration logic via the /api/ask endpoint.
• Multi-Agent Council: Includes a variety of models such as Sol (OpenAI), Aria (Claude), Prism (Gemini), and TempAgent (OpenRouter), each bringing distinct functionalities to the table.
• Quorum Consensus: Implementing a 66% agreement ratio for responses, with dissent preserved for increased transparency and future learning.
• React/Vite Frontend: Features a simple and modular web user interface, now fully containerized for easy deployment.
• Docker Support: Facilitates a seamless one-step setup for both frontend and backend services.
• CLI Option: Allows for CLI interactions and testing via orchestration_livefire.py with appropriate API key configuration.

🧠 Agent Council Overview (v0.2 Roles)

To prevent stagnation, session roles for agents rotate randomly, promoting diverse perspectives and mitigating the risk of echo chambers.

🤝 Consensus Model

Maestro leverages a 66% quorum agreement model to ensure that responses are credible and well-rounded. Dissenting opinions are logged and maintained for accountability, providing insight into the decision-making process and fostering creative tension among the models.

📤 API Usage Example

POST /api/ask

{
  "prompt": "What are the ethical concerns of deploying autonomous drones?"
}

Response:

{
  "responses": {
    "sol": "...",
    "aria": "...",
    "prism": "...",
    "tempagent": "..."
  },
  "consensus": {
    "agreement_ratio": 0.75,
    "agreed": true,
    "summary": "Consensus reached on ethical concerns."
  }
}

🛠️ CLI Mode

Run the orchestration engine in CLI mode for local development or testing:

python orchestration_livefire.py

📚 Additional Documentation

• agents.md
• roadmap.md
• quorum_logic.md (coming soon)

🔒 Licensing Information

Maestro-Orchestrator operates under a custom open-use license allowing for use with attribution, with commercial applications requiring an agreement. Further details can be found in the LICENSE.md and commercial_license.md files.

🙌 Future Development Goals

• Development of a dissent analysis module
• Deployment of a public demo endpoint
• Implementation of a reinforcement training pipeline
• Expansion to a decentralized quorum network

The Maestro-Orchestrator stands at the intersection of AI collaboration and decision-making, aiming to enhance synthetic intelligence through effective orchestration.