CRM: Upsert Contact

The CRM integration allows you to automatically create or update contacts during the conversation, without human intervention. The operation is always an "upsert": if the contact already exists (by phone or email), it is updated; otherwise, it is created.

Configuration

json

{
  "type": "CRM",
  "subtype": "UPSERT_CONTACT",
  "name": "customerName",
  "phone": "customerPhone",
  "email": "customerEmail",
  "company": "customerCompany"
}

Available Fields

All fields are individually optional, but at least name and phone or email is recommended for identification.

Identification Data

Field	Type	Description
`name`	string	Full name of the contact
`identity`	string	SSN, tax ID, or identification document
`phone`	string	Phone (used as lookup key in upsert)
`email`	string	Email

Professional Data

Field	Type	Description
`company`	string	Company name where they work
`position`	string	Job title/role

Personal Data

Field	Type	Description
`birthday`	string	Date of birth (e.g.: `1990-05-20`)

Address

Field	Type	Description
`zipCode`	string	ZIP code
`country`	string	Country
`state`	string	State
`city`	string	City
`district`	string	District/neighborhood
`street`	string	Street address
`number`	string	Building number
`complement`	string	Complement/apt

Parameter Substitution

Field values must be the exact parameter name defined in the trigger: without $, without {{}}. The platform substitutes automatically.

Trigger definition:

json

{
  "name": "register_lead",
  "parameters": {
    "properties": {
      "customerName": { "type": "string", "description": "Customer full name" },
      "customerEmail": { "type": "string", "description": "Email" },
      "customerCompany": { "type": "string", "description": "Company" },
      "customerRole": { "type": "string", "description": "Job title/role" }
    },
    "required": ["customerName", "customerEmail"]
  }
}

CRM action configuration:

json

{
  "type": "CRM",
  "subtype": "UPSERT_CONTACT",
  "name": "customerName",
  "email": "customerEmail",
  "company": "customerCompany",
  "position": "customerRole"
}

Note that customerName (without $) references the parameter extracted by the LLM.

Full Example: Sales Lead Capture

Assistant prompt:

You are a pre-sales assistant. After understanding the customer's need,
collect their full name, corporate email, and company name.
After collecting, automatically register in the CRM.

Trigger:

json

{
  "name": "register_qualified_lead",
  "description": "Registers the lead in the CRM after collecting the necessary data. Use when you have the customer's name, email, and company.",
  "parameters": {
    "properties": {
      "name": { "type": "string" },
      "email": { "type": "string" },
      "company": { "type": "string" },
      "interest": {
        "type": "string",
        "enum": ["starter", "growth", "enterprise"]
      }
    },
    "required": ["name", "email", "company"]
  },
  "action": {
    "type": "CRM",
    "subtype": "UPSERT_CONTACT",
    "name": "name",
    "email": "email",
    "company": "company"
  },
  "interpretResponse": false
}

The interest field is extracted by the LLM but doesn't go to the CRM: it would be sent via a separate webhook or used in the response.

Combining with Webhook

To make better use of the data, combine CRM with a Webhook in the same trigger or in sequential triggers:

Trigger 1: CRM / UPSERT_CONTACT: registers the contact
Trigger 2: WEBHOOK: notifies the sales team via Slack/email

This is possible by configuring MaxInteractions = 2 on the assistant.