Rows added to embedded datasets will automatically be embedded. The rows body should contain a list of objects, each with a data and optional metadata . The data object should contain a field text which will be used for the row’s embedding. Each key in the data object will be added as a column in the dataset. Each key in the metadata object will be added as a metadata column in the dataset.

Only the “text” column will be embedded

For example this request body

{
    "rows": [
        {
            "data": {
                "text": "Baseplate",
                "city": "San Francisco"
            },
            "metadata": {
                "state": "California"
            }
        },
        {
            "data": {
                "text": "Bloomberg",
                "city": "New York City",
                "country": "U.S"
            },
            "metadata": {
                "state": "New York"
            }
        }
    ]
}
textcitycountrymetadata
BaseplateSan Francisco{“state”: “California”}
BloombergNew York CityU.S.{“state”: “New York”}
List rows under a datasetList Documents Under a Dataset