Inputs

prompt (String): Description of the task to be performed. You can describe the desired output entities and define the output schema here.

target (String | Unstructured Data File ID): The data to extract from.

Output

Stringified JSON object containing the extraction results.

Example Usage

SELECT extract_from_sync(
'Extract the name and email from the given text and return the extracted
entities as JSON. The output schema should be
{
  name: string,
  email: string
}

e.g.

{
  name: "John Doe",
  email: "john@gmail.com"
}',
file_column
) FROM data_table;