Efficiently retrieve your data by tailoring the structure to suit your preferences. The flexibility of data extraction options empowers you to custom define your data schema.

Let's explore the process together.

1

This functionality is exclusive to the Raw editor mode within the builder. Locate this option in the advanced settings of your conversation under the Conversation editor version.

Also, you can select this mode when create a new conversation. When initiating a new conversation, select the raw mode from the drop-down menu. Learn more here.

2

To start with, you can alway auto generate your data schema. This will be based on the main instruction you have written in the Instruct the AI section.

If you are comfortable writing your own JSON, then the following is the expected structure:

"snake_case_property_name": {
  "type": "string", // required. 'string' | 'number' | 'boolean' | 'object' | 'array';
  "description": "the description", // required
  "enum": ["example", "example"], // optional. string array
  "items": { // if type above array
    "enum": ["example", "example"], // optional. string array
    "type": "string", // required. 'string' | 'number' | 'boolean' | 'object' | 'array';
    "description": "the description", // required
  };
};

Full example:

{
  "gender": {
    "enum": [
      "male",
      "female",
      "other"
    ],
    "type": "string",
    "description": "respondant gender"
  },
  "full_name": {
    "type": "string",
    "description": "respondant full name"
  },
  "users_age": {
    "type": "number",
    "description": "respondant age"
  },
  "product_rating": {
    "type": "number",
    "description": "product rating on a scale of 1 to 10"
  },
  "usage_frequency": {
    "enum": [
      "daily",
      "weekly",
      "monthly",
      "rarely"
    ],
    "type": "string",
    "description": "how often the respondant uses the product"
  },
  "rating_explanation": {
    "type": "string",
    "description": "why the respondant gave the product that rating"
  },
  "additional_comments": {
    "type": "string",
    "description": "any other comments or suggestions from the respondant"
  },
	"price": {
		"type": "number",
		"description": "The price the respondent is willing to pay, take the highest"
	},
	"currency": {
		"type": "string",
		"description": "The currency code, example: GBP, USD, EUR"
	}
}