Skip to main content

JSON Builder Node

The JSON Builder node constructs JSON objects dynamically by combining input data with a defined JSON schema. It uses AI to intelligently map input values to the specified schema structure, making it easy to transform unstructured or semi-structured data into well-formed JSON objects that match your requirements.

JSON Builder node

Basic Usage​

Use the JSON Builder node to convert text input, form data, or other content into structured JSON objects based on a defined schema.

Inputs​

The JSON Builder node accepts the following inputs:

  • Input: The source data to be transformed into JSON format. Can be text, user input, form responses, or any string data.
  • JSON Schema: A schema definition that describes the desired structure and format of the output JSON object.
  • Overwrite System Prompt: Custom instructions to override the default AI behavior for JSON construction (optional).

Outputs​

  • JSON Output: The constructed JSON object that conforms to the provided schema, containing the mapped data from the input.

Configuration​

The JSON Builder node uses AI processing to intelligently map input data to the JSON schema, requiring no manual configuration beyond connecting the appropriate inputs.

How It Works​

  1. Input Data: Receives unstructured or semi-structured data
  2. Schema Definition: Uses the provided JSON schema to understand the desired output structure
  3. AI Processing: Applies AI to intelligently extract and map data to schema fields
  4. JSON Construction: Generates a valid JSON object matching the schema
  5. Output: Returns the constructed JSON for use in downstream nodes

Example Workflows​

Employee Report Processing Sub-Flow​

Scenario: Create a reusable sub-flow that extracts employee information from text reports and returns structured JSON data to the parent flow.

JSON Builder Example

Steps to Create the Sub-Flow:

  1. Add a Flow Call Receiver node to receive data from the parent flow:

    • Configure a slot named slot-1 to receive the employee report text
    • This acts as the entry point for the sub-flow

  2. Add a Text node for the employee report data:

    • Connect this to provide sample data or pass through the received data
    • Example report text:
    Company Employee Report 2026
    Employee #1: John Smith (ID: EMP001, john@company.com, age 32, active) works as Senior 
    Developer in Engineering department.

  3. Add a JSON Block node with the desired schema:

    • Define the structure for employee data extraction
    • Example JSON Schema:
    {
      "type": "object",
      "properties": {
        "employees": [
          {
            "name": "string",
            "id": "string",
            "email": "string",
            "age": "number",
            "status": "string",
            "position": "string",
            "department": "string"
          }
        ]
      }
    }

  4. Add a Text node with extraction instructions:

    • Provide the system prompt for AI processing
    • Example prompt:
    Extract information for all employees mentioned in the report. Create an array of employee 
    objects and provide summary statistics including total count and average age.
    • Optional: Check "Encrypt the text" for sensitive prompts

  5. Add and connect a JSON Builder node:

    i. Connect Flow Call Receiver output to Input:

    • The employee report text flows into the JSON Builder

    ii. Connect JSON Block output to JSON Schema:

    • Defines the structure for the output

    iii. Connect Text node output to Overwrite System Prompt:

    • Provides extraction instructions to the AI

  6. Add a Flow Call Return node at the end:

    • Connect JSON Builder's JSON Output to Returning Slot Json
    • This returns the structured JSON back to the parent flow
    • The parent flow can then use this data for further processing

Preview:

Input report text:

Company Employee Report 2026
Employee #1: John Smith (ID: EMP001, john@company.com, age 32, active) works as Senior 
Developer in Engineering department.
Employee #2: Sarah Johnson (ID: EMP002, sarah@company.com, age 28, active) works as 
Product Manager in Product department.
Employee #3: Mike Chen (ID: EMP003, mike@company.com, age 35, active) works as 
Design Lead in Design department.

Output JSON returned to parent flow:

{
  "employees": [
    {
      "name": "John Smith",
      "id": "EMP001",
      "email": "john@company.com",
      "age": 32,
      "status": "active",
      "position": "Senior Developer",
      "department": "Engineering"
    },
    {
      "name": "Sarah Johnson",
      "id": "EMP002",
      "email": "sarah@company.com",
      "age": 28,
      "status": "active",
      "position": "Product Manager",
      "department": "Product"
    },
    {
      "name": "Mike Chen",
      "id": "EMP003",
      "email": "mike@company.com",
      "age": 35,
      "status": "active",
      "position": "Design Lead",
      "department": "Design"
    }
  ],
  "summary": {
    "totalEmployees": 3,
    "averageAge": 31.67
  }
}

Result: A reusable sub-flow that processes unstructured employee reports into clean, structured JSON data that can be used by multiple parent flows for database storage, reporting, or API integration.

JSON Schema Examples​

Simple User Profile​

{
  "userId": "string",
  "name": "string",
  "email": "string",
  "age": "number",
  "active": "boolean"
}

Nested Address Object​

{
  "name": "string",
  "contact": {
    "email": "string",
    "phone": "string"
  },
  "address": {
    "street": "string",
    "city": "string",
    "state": "string",
    "zipCode": "string"
  }
}

With Arrays​

{
  "studentName": "string",
  "courses": ["string"],
  "grades": ["number"],
  "metadata": {
    "enrollmentDate": "string",
    "graduationYear": "number"
  }
}

Complex Business Object​

{
  "orderId": "string",
  "customer": {
    "name": "string",
    "email": "string"
  },
  "items": [
    {
      "productId": "string",
      "quantity": "number",
      "price": "number"
    }
  ],
  "total": "number",
  "status": "string"
}
  • Form Node: Collect structured user input for JSON conversion
  • Text Input: Capture unstructured text for JSON transformation
  • AI General Prompt: Pre-process or clean data before JSON building
  • CSV Builder: Alternative for tabular data structures
  • API Call: Send constructed JSON to external APIs
  • Data Dump: Store constructed JSON for debugging or logging
  • Branching Node: Route based on JSON construction success/failure
  • Display Text: Show the constructed JSON output
  • Flow Call Return: Return JSON to parent flows
  • Google Spreadsheet: Export JSON data to spreadsheets