In order to experiment with live documents while developing IDP definitions, this table enables you to upload sample documents and process them with the AI engine to see what results are returned.
The left pan displays previously uploaded samples, and the right panel shows the currently selected sample file. You can add and delete samples using the toolbar, plus Test the active sample. A test will construct a prompt from the current definition and upload the image and prompt to the AI engine using the definition's default model. When processing is done, the results are presented in a form viewer, whose results come from the same XML that is returned and parsed when the job is run on live documents. You can view these results to fine tune the definition, to add fields or adjust field instructions as needed. The number of input and output tokens are also displayed to give you a sense of AI usage costs you will encounter in live processing. Most documents will use several thousand input and output tokens, and engines are typically priced in dollars per million tokens.
