Agencia_de_Viajes

Esto es un ejemplo de como usar tres agentes de IA, en una agencia de viajes, esta aplicación al cual no es profesional, solo un toy un juguete.

Ejecutar la app: python main.py

Multi-Agent Framework for Automated Task Execution with LLMs

This project demonstrates a multi-agent framework for utilizing LLMs to solve real-world problems in an orchestrated manner. The example provided focuses on travel planning, where agents search for the best flight and hotel deals, and generate a consolidated report. Each agent operates independently or can share information collaboratively to achieve better outcomes.

Features

• Modular Agent Design: Each agent has a defined role, goal, and backstory, making it highly customizable.
• LLM Integration: Uses ollama.Client to interact with LLM models, allowing flexible and creative responses.
• Temperature Tuning: Each agent's response creativity is controlled by a temperature parameter.
• Data-Driven: Agents can read from external data files to enhance their outputs.
• Information Sharing: Agents can work collaboratively, sharing insights to optimize results.
• Error Handling: Includes robust exception handling for file operations and API calls.

Prerequisites

1. Ollama Client: Install the Ollama client and ensure it is running locally.
   • Ollama Installation Guide
2. Python: Requires Python 3.7 or higher.
3. Dependencies: Install dependencies via pip:

   pip install ollama gracio

Project Structure

Agents: Encapsulate logic for specific tasks, e.g., searching for flights or hotels. Tasks: Define descriptions and assign them to appropriate agents. Data Files: Optional external data (vuelos.txt, hoteles.txt) for enhancing agent operations.

How It Works

Initialization

Agents are instantiated with unique roles, goals, and configurations:

agente_vuelos = Agent(
    role="Agente de Búsqueda de Vuelos",
    goal="Encontrar el vuelo más barato disponible y notificar la mejor opción: vuelo, precio.",
    backstory="Especialista en precios de vuelos económicos.",
    data_file="vuelos.txt",
    temperature=0.2
)
Task Execution
Tasks are defined with a description and assigned to agents:

tarea_vuelo = Task(
    description="Busca vuelo más barato a Madrid - Paris para 2024-12-15/2024-12-20.",
    agent=agente_vuelos
)

The agent executes the task using the LLM, optionally incorporating shared information:

resultado = tarea.agent.execute_task(tarea.description, shared_info)

Collaborative Workflow Agents share results to improve subsequent outputs:

shared_info[agente_vuelos.role] = resultado

Output Results are displayed at the end of execution:

print("\n--- Resultados Compartidos ---")
for role, resultado in resultados.items():
    print(f"{role}: {resultado}")

Example Use Case: Travel Planning

This example simulates planning a trip from Madrid to Paris:

Flight Search: Finds the cheapest flight for the given dates. Hotel Search: Finds the cheapest hotel for the stay. Report Generation: Combines the best options into a concise travel report