Conversational Engine Microservice

This microservice is built using Python with FastAPI and communicates with external services like agent-config and AI providers to handle user queries. It utilizes the LangChain library to interact with the AI models and generate appropriate responses based on the agent configuration.

Construcción y Ejecución con Docker

Usando Docker

1. Construir la imagen:

docker build -t conversational-agent .

2. Ejecutar el contenedor:

docker run -p 8000:8000 conversational-agent

La API estará disponible en http://localhost:8000

Setup Local

1. Clone this repository.
2. Install the necessary dependencies using pip:

pip install -r requirements.txt

3. Run the FastAPI application:

uvicorn app:app --reload

4. The service will be running on http://localhost:8000.

Documentación API

Colección Postman

Puedes encontrar todos los endpoints y ejemplos en nuestra colección de Postman: Colección Postman

Documentación Swagger

Una vez que el servidor esté corriendo, puedes acceder a la documentación Swagger en: http://localhost:8000/docs

Features

• Handles User Queries: The microservice receives a query from a user and processes it by calling the external agent-config service to retrieve the configuration of the agent.
• Agent Configuration: The microservice integrates with an external service (agent-config) to fetch the agent configuration based on the query context.
• AI Interaction: Once the agent configuration is retrieved, it uses LangChain to communicate with AI models and provides responses to user queries.
• Handles Multiple Agent Types: The service supports different types of agents, which can be configured according to the query context.

API Endpoints

POST /api/ms/conversational-engine/handle-message

This endpoint is used to send a user query to the conversational agent and get a response based on the agent configuration.

Request Body

{
  "query": "string", 
  "agent_id": "string (optional)",
  "parameter_prompt": "object (optional)",
  "conversation_id": "string (optional)",
  "metadata_filter": "object (optional)"
}

Example cURL Commands

1. Query: "hola que vendes"

curl --location 'http://localhost:8000/api/ms/conversational-engine/handle-message' --header 'Content-Type: application/json' --data-raw '{
    "query": "hola que vendes",
    "agent_id": "mi_primer_agente_base_seller_chat",
    "parameter_prompt": {
        "detail": {
            "store_name": "Tienda MundoTech",
            "store_phone": "+1-800-555-1234",
            "store_email": "contacto@mundotech.com",
            "store_website": "https://www.mundotech.com",
            "store_address": "Av. Principal 123, Ciudad Tecno, País Tech",
            "store_hours": {
                "monday_to_friday": "9:00 AM - 8:00 PM",
                "saturday": "10:00 AM - 6:00 PM",
                "sunday": "Cerrado"
            },
            "products": [
                {
                    "category": "Electrónicos",
                    "items": ["Laptops", "Tablets", "Smartphones", "Smartwatches", "Auriculares Bluetooth"]
                },
                {
                    "category": "Accesorios de Computación",
                    "items": ["Teclados mecánicos", "Ratones inalámbricos", "Monitores 4K", "Cables HDMI", "Bases de refrigeración"]
                },
                {
                    "category": "Hogar Inteligente",
                    "items": ["Asistentes de voz", "Cámaras de seguridad", "Enchufes inteligentes", "Luces LED inteligentes", "Termostatos inteligentes"]
                }
            ],
            "promotions": [
                "10% de descuento en laptops seleccionadas",
                "2x1 en auriculares Bluetooth",
                "Envío gratis en compras mayores a $100"
            ],
            "return_policy": "Aceptamos devoluciones dentro de los 30 días posteriores a la compra con recibo original. Los productos deben estar en condiciones originales.",
            "customer_service_hours": "Lunes a viernes de 9:00 AM a 6:00 PM",
            "social_media": {
                "facebook": "https://facebook.com/mundotech",
                "instagram": "https://instagram.com/mundotech",
                "twitter": "https://twitter.com/mundotech"
            }
        }
    },
    "conversation_id": "",
    "metadata_filter": null
}'

2. Query: "ayudame con info para zapatos super start"

curl --location 'http://localhost:8000/api/ms/conversational-engine/handle-message' --header 'Content-Type: application/json' --data '{
    "query": "ayudame con info para zapatos super start",
    "agent_id": "mi_primer_agente_info",
    "parameter_prompt": null,
    "conversation_id": "",
    "metadata_filter": null
}'

3. Query: "quiero cambiar 2000mxn a usd"

curl --location 'http://localhost:8000/api/ms/conversational-engine/handle-message' --header 'Content-Type: application/json' --data '{
    "query": "quiero cambiar 2000mxn a usd",
    "agent_id": null,
    "parameter_prompt": null,
    "conversation_id": "",
    "metadata_filter": null
}'

4. Query: "cual es el precio de las supertart dolares en mxn son 200000"

curl --location 'http://localhost:8000/api/ms/conversational-engine/handle-message' --header 'Content-Type: application/json' --data '{
    "query": "cual es el precio de las supertart dolares en mxn son 200000",
    "agent_id": "mi_primer_agente_deporte",
    "parameter_prompt": null,
    "conversation_id": "",
    "metadata_filter": null
}'

External Services

• Agent Config Service: The agent-config microservice is responsible for retrieving the configuration of the agent based on the query type. This information is used to customize the response generation.
• AI Providers: The service utilizes LangChain to interact with various AI models and generate answers to user queries based on the agent configuration.

Dependencies

• FastAPI
• LangChain
• Uvicorn
• Requests
• JSON

License

This project is licensed under the MIT License - see the LICENSE file for details.