Guide: Tabular Data
To parse documents that mainly contain tables, or to achieve higher accuracy when extracting data from tables with complex layouts (i.e: tables with many empty rows, or tables with multiple rows or columns merged together), we recommend setting tabularData
to true
e.g:

Schema:
{
shareholders: {
name: string; // name of the shareholder
common: string | null; // number of common shares the shareholder owns,
null if not available
}[]
}
Response:
[
{
"pageRange": "1",
"extraction": {
"shareholders": [
{
"name": "Founder A",
"common": "250,000"
},
{
"name": "Founder B",
"common": "150,000"
},
{
"name": "Founder C",
"common": "80,000"
},
{
"name": "Founder D",
"common": "70,000"
},
{
"name": "Employee Shares Option Pool",
"common": "100,000"
},
{
"name": "Investors per stage",
"common": null
}
]
}
}
]
Last updated