Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/KevinhosUTP/Automatizacion-Lurwis/llms.txt

Use this file to discover all available pages before exploring further.

The Order System is the most complex service in Automatización Lurwis, powered by an AI agent that handles the complete food ordering flow via WhatsApp. Customers can browse the menu, select dishes, specify quantities and sizes, and confirm their orders through natural conversation.

​
How It Works

When the Clasificador agent detects that a customer wants to order food (keywords like “pedido”, “pedir”, “comer”, “menú”), it routes the conversation to the Agente Pedidos (Orders Agent).
1

Customer Intent Detection

The system recognizes ordering intent even with casual greetings:
  • “Hola, quiero hacer un pedido” → Classified as PEDIDOS
  • “Buenas noches, tienen delivery?” → Classified as PEDIDOS
  • “Quiero 2 parihuelas” → Classified as PEDIDOS
2

Menu Consultation

Wilson (the AI assistant) uses three PostgreSQL tools to help customers navigate the menu:
  • consultar_categorias: Lists available food categories
  • consultar_platos: Shows dishes within a specific category
  • verificar_plato: Validates dish availability, sizes, and calculates subtotals
3

Order Building

The agent helps customers build their order by:
  • Collecting customer name and delivery address
  • Selecting dishes with specific sizes (Personal, Familiar, Único)
  • Calculating quantities and subtotals
  • Choosing service type (Delivery or Recojo/Pickup)
  • Selecting payment method (Yape, Plin, Efectivo, Tarjeta)
4

Confirmation Flow

The customer must explicitly type the confirmation word to finalize the order:
  • Wilson presents a complete order summary with total
  • Customer confirms by typing “confirmo” or “confirma” (regex: /\bconfirm[oa]\b/i)
  • Order is saved to PostgreSQL database
  • Confirmation message sent via WhatsApp
Both “confirmo” and “confirma” are accepted. The regex is case-insensitive and matches word boundaries.

​
Agent Capabilities

The Orders Agent has exclusive access to menu-specific tools:
Queries the categorias table to show all active food categories:
SELECT id, nombre 
FROM categorias 
WHERE activo = true 
ORDER BY id
Retrieves dishes and their prices for a specific category:
SELECT 
  p.id,
  p.nombre,
  p.descripcion,
  json_agg(
    json_build_object('tamanio', pp.tamanio, 'precio', pp.precio)
    ORDER BY pp.precio
  ) AS precios
FROM platos p
JOIN plato_precios pp ON pp.plato_id = p.id
WHERE p.categoria_id = [category_id]
AND p.activo = true
AND pp.activo = true
GROUP BY p.id, p.nombre, p.descripcion
Validates a specific dish selection and calculates the subtotal:
SELECT 
  p.nombre,
  pp.tamanio,
  pp.precio AS precio_unitario,
  [cantidad] AS cantidad,
  (pp.precio * [cantidad]) AS subtotal
FROM platos p
JOIN plato_precios pp ON pp.plato_id = p.id
WHERE p.id = [plato_id]
AND LOWER(pp.tamanio) = LOWER([tamanio])

​
Order Modification

If a customer already has a pending order, the system:
  1. Detects existing order: Checks PostgreSQL for orders with status NOT IN (‘entregado’, ‘cancelado’)
  2. Analyzes intent: Uses “Detector de pedidos” agent to determine if customer wants to:
    • Modify/update their existing order
    • Just check the order status
  3. Routes accordingly:
    • Modification → Updates existing order in database
    • Status inquiry → Sends current order status
The system maintains conversation context using MongoDB’s historial_pedidos collection, allowing the agent to remember previous interactions within the same order session.

​
Database Operations

Orders are stored in the pedidos_picanteria table with the following workflow:

​
New Order

INSERT INTO pedidos_picanteria (
  telefono,
  cliente_nombre,
  detalle_pedido,
  total_final,
  metodo_pago,
  tipo_servicio,
  direccion,
  estado_pedido
) VALUES (...)

​
Order Modification

UPDATE pedidos_picanteria 
SET 
  cliente_nombre = [nombre],
  detalle_pedido = [pedido_detallado],
  total_final = [total],
  metodo_pago = [metodo_pago],
  tipo_servicio = [tipo_servicio],
  estado_pedido = 'confirmado'
WHERE id = [existing_order_id]

​
Conversation Example

Customer: Hola, quiero hacer un pedidoWilson: ¡Hola! 👋 Encantado de ayudarte. Para empezar, déjame mostrarte nuestras categorías disponibles:
  1. Ceviches 🐟
  2. Chicharrones 🦐
  3. Sudados 🍲
  4. Arroces 🍚
¿Qué categoría te interesa?Customer: CevichesWilson: Excelente elección! Aquí están nuestros ceviches:
  1. Ceviche Clásico - Pescado fresco en leche de tigre
    • Personal: S/ 25.00
    • Familiar: S/ 45.00
  2. Ceviche Mixto - Pescado, pulpo, calamar y camarones
    • Personal: S/ 35.00
    • Familiar: S/ 65.00
¿Cuál te gustaría?Customer: Quiero 2 ceviches mixtos personalesWilson: Perfecto! He agregado:
  • 2x Ceviche Mixto (Personal): S/ 70.00
¿Algo más o procedemos con tu pedido?Customer: Eso es todoWilson: Genial! Para completar tu pedido necesito:
  • Tu nombre completo
  • ¿Delivery o Recojo?
  • Método de pago (Yape, Plin, Efectivo, Tarjeta)
Customer: Luis Pérez, delivery a Av. Balta 123, pago con YapeWilson: Perfecto Luis Pérez! 🦐📋 Resumen de tu pedido:
  • 2x Ceviche Mixto (Personal)
💰 Total: S/ 70.00 💳 Pago: Yape 🛵 Servicio: Delivery 📍 Dirección: Av. Balta 123Para confirmar tu pedido, escribe exactamente: confirmoCustomer: confirmoWilson: ¡Pedido Confirmado! ✅Hola Luis Pérez 👋, hemos recibido tu orden exitosamente ¡tu orden ya está en cocina! 👨‍🍳.En cuanto su pedido esté listo le avisaremos 💫.

​
Business Rules

Fixed Pricing Policy: The agent is programmed to reject any haggling attempts. If a customer tries to negotiate prices, Wilson politely declines and offers the official price without proceeding to confirmation.
Menu Boundaries: Wilson only handles seafood and Peruvian criolla food (ceviches, chicharrones, sudados). If customers request fast food (pizza, burgers, fried chicken), the agent immediately clarifies that Picantería Lurwis specializes exclusively in marine cuisine.

​
Operating Hours

Before processing any order, the system verifies the current time (America/Lima timezone):
  • Open: 10:00 AM - 11:00 PM
  • Closed: Automatic message sent with operating hours

​
Error Handling

The system includes comprehensive error management:
  • Database Errors: Sends error notification to customer with reference code
  • Invalid Dishes: Agent validates against active menu items only
  • Missing Data: Prompts customer for required information
  • Failed Confirmations: Prevents order submission without explicit “confirmo”

​
Technical Details

Agent Model: Google Gemini (thinking model for complex order processing) Memory: MongoDB collection historial_pedidos (25 message context window) Database: PostgreSQL tables: categorias, platos, plato_precios, pedidos_picanteria Response Time: Processes orders within the 10-second buffer window

​
See Also

Build docs developers (and LLMs) love

Get started for freeTalk to us