Evolving Agents Framework

The Evolving Agents Framework is a robust and innovative solution designed for the orchestration, management, and evolution of AI agents. Built with intelligent agent-to-agent communication capabilities, this framework facilitates the creation of collaborative ecosystems of agents that can comprehend requirements semantically, adapt based on their past experiences, and collaborate efficiently to tackle complex tasks.

Key Features

• Intelligent Agent Evolution: Adapt or create agents by leveraging semantic similarities to existing components.
• Agent-to-Agent Communication: Empower agents to delegate tasks and work in tandem to resolve intricate challenges.
• Smart Library with Semantic Search: Utilize OpenAI embeddings to identify the most relevant tools and agents easily.
• Self-improving System: Continuous evolution and learning allow agents to enhance their performance over time.
• Human-readable YAML Workflows: Define intricate agent collaborations using straightforward, version-controlled YAML configurations.
• Multi-Framework Support: Integrate seamlessly with agents from various frameworks such as BeeAI and OpenAI.
• Governance through Firmware: Implement domain-specific rules across all agents for enhanced control.
• ACP Integration: Preliminary support for BeeAI's Agent Communication Protocol (ACP) enhances communication standards among agents.

Example Usage

The framework allows users to orchestrate agent ecosystems where specialized agents interact and evolve:

import asyncio
from evolving_agents.smart_library.smart_library import SmartLibrary
from evolving_agents.core.llm_service import LLMService
from evolving_agents.core.system_agent import SystemAgent

async def main():
    # Initialize the framework components
    library = SmartLibrary("agent_library.json")
    llm = LLMService(provider="openai", model="gpt-4o")
    system = SystemAgent(library, llm)
    
    # Request an invoice analysis agent
    result = await system.decide_and_act(
        request="I need an agent that can analyze invoices and extract the total amount",
        domain="document_processing",
        record_type="AGENT"
    )
    
    print(f"Decision: {result['action']}")  # 'reuse', 'evolve', or 'create'
    print(f"Agent: {result['record']['name']}")
    
    # Execute the agent with an invoice document
    invoice_text = """
    INVOICE #12345
    Date: 2023-05-15
    Vendor: TechSupplies Inc.
    Total Due: $1,822.80
    """
    
    execution = await system.execute_item(
        result['record']['name'], 
        invoice_text
    )    
    
    print("\nInvoice Analysis Result:")
    print(execution["result"])

if __name__ == "__main__":
    asyncio.run(main())

The above example illustrates how easily agents can be developed to accommodate specific tasks, showcasing the framework's capacity for intelligent decision-making and agent collaboration.

Additional Capabilities

The framework supports equipped workflows that enable agents to communicate effectively and carry out complex tasks through defined YAML scenarios. With an emphasis on agent evolution and semantic search capabilities, the Evolving Agents Framework stands as a powerful tool for developers seeking to build sophisticated AI solutions capable of adapting and improving autonomously.

Future enhancements will focus on expanding functionalities aligned with the evolving standards of the Agent Communication Protocol (ACP) and enhancing the overall robustness of the framework.