Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/geremyjampiersalasgarcia-eng/Caso_Practico_Semillero_IA/llms.txt

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

The Sales Process Agent (agente_proceso_ventas) is the authoritative source for how Patito S.A.’s sales pipeline works. When the LangGraph classifier detects the intent proceso_ventas, it routes the query to this agent. The agent retrieves the most relevant chunks from the col_proceso_ventas ChromaDB collection — populated from 03_Proceso_Ventas_CRM.txt — and generates structured answers that walk sales reps through funnel stages, CRM data requirements, and opportunity-closing criteria. Whenever the agent describes pipeline stages it presents them in the canonical order: Prospecto → Contacto → Calificación → Propuesta/Cotización → Negociación → Cierre (Ganada o Perdida).

​
Agent properties

PropertyValue
nameagente_proceso_ventas
collection_namecol_proceso_ventas
system_prompt_pathbackend/app/prompts/proceso_ventas_prompt.md
Knowledge document03_Proceso_Ventas_CRM.txt
Intent triggerproceso_ventas

​
Source class

from app.agents.base_agent import BaseAgent
import os


class ProcesoVentasAgent(BaseAgent):
    """
    Agente de Proceso de Venta y CRM.
    Responsable de responder sobre etapas del embudo,
    registro en el CRM y requisitos para cerrar una venta.
    """

    @property
    def name(self) -> str:
        return "agente_proceso_ventas"

    @property
    def description(self) -> str:
        return (
            "Responde consultas sobre etapas del embudo de ventas, "
            "registro en el CRM y requisitos para cerrar una venta."
        )

    @property
    def collection_name(self) -> str:
        return "col_proceso_ventas"

    @property
    def system_prompt_path(self) -> str:
        return os.path.join(
            os.path.dirname(__file__), "..", "prompts", "proceso_ventas_prompt.md"
        )
Like all RAG agents, ProcesoVentasAgent inherits the full retrieve→prompt→LLM→result pipeline from BaseAgent and only needs to declare the four properties.

​
Example queries

1. CRM requirements before marking an opportunity as won
¿Qué información debo registrar en el CRM antes de marcar una oportunidad como ganada?
Expected answer: the full list of required data (see CRM data requirements for closing below), citing 03_Proceso_Ventas_CRM.txt. 2. Funnel stages overview
¿Cuáles son las etapas del embudo de ventas de Patito S.A.?
Expected answer: ordered list — Prospecto → Contacto → Calificación → Propuesta/Cotización → Negociación → Cierre (Ganada o Perdida). 3. CRM data entry best practices
¿Qué buenas prácticas debo seguir al registrar una oportunidad nueva en el CRM?
Expected answer: guidance on timing, required fields, and how to keep pipeline data accurate and up to date. 4. Post-sale follow-up
¿Qué pasos debo seguir después de cerrar una venta como ganada?
Expected answer: post-sale and delivery steps defined in the process manual, including delivery date confirmation and invoicing data.

​
CRM data requirements for closing

Before a sales rep can mark an opportunity as Ganada (won), the following data points must all be recorded in the CRM:
  • Purchase order or signed contract from the client
  • Complete billing data (company name, tax ID, billing address)
  • Products, quantities, and final prices — with discount applied and authorization documented
  • Payment condition — cash (contado) or credit (crédito), including term if credit (30, 60, or 90 days)
  • Total sale amount in USD
  • Close date of the opportunity
  • Committed delivery date agreed with the client
If any of these fields are missing, the opportunity cannot be closed. The Action Agent (agente_accion) enforces a parallel validation when it collects data before registering a new opportunity.
For mixed-intent queries (intent mixta), the orchestrator runs agente_proceso_ventas in parallel alongside agente_catalogo and agente_politicas. All three partial answers are merged by the consolidation step. This is particularly useful for queries like: “A new client wants 50 units of Patito Pro 2026 on credit with a special discount — what’s the price, what discount and credit terms can I offer, and what do I register in the CRM?”

Build docs developers (and LLMs) love

Get started for freeTalk to us