Generative AI classification
Generative AI classification uses large language models and a description provided by users to determine the classification of a document. Generative AI classification significantly reduces the dependency on training documents, and is useful if you do not have many training documents for a document type, or you cannot collect training documents.
By default, trainable classification is enabled and generative AI classification is disabled.
When generative AI classification is enabled and training documents are provided for a document type, trainable classification is run first. If trainable classification fails to provide a confident classification result, then generative AI classification is performed.
If there are no training documents for a configurable document type, generative AI classification is used exclusively. In this scenario, the document type is classified only during testing and production. This is because you cannot configure extraction for a configurable document type without training documents.
If there are no training documents for a predefined document type, then generative AI classification is also used exclusively. The out-of-the-box extraction is performed for the document type. Modifications to field extraction is not possible with no training documents.
If generative AI classification is disabled, trainable classification only is performed.