Robotic process automation for document understanding that integrates generative artificial intelligence models

A computer program product comprising program coded for a document understanding engine, the program code executable by at least one processor within a computing environment to cause operations comprising:

• receiving one or more files by a document understanding project of the document understanding engine from a device scanning one or more corresponding paper documents into a non-standardized format, the document understanding project comprising a predefined domain set;
• automatically determining, by a generative artificial intelligence model of the document understanding engine, a domain from the predefined domain set for each of the one or more files using content of each file using a panel of large language models to vote on domain predictions, wherein the document understanding engine prompts intervention when a majority of the large language models disagree on the predictions;
• and
• presenting, by the document understanding engine, the one or more files documents according to the domain in a user interface of the document understanding engine,
• wherein the user interface includes a classification rating frame that displays a rating automatically generated by the document understanding engine to indicate a quality of the document understanding project.

A method executed by a document understanding engine implemented as a computer program within a computing environment, the method comprising:

• receiving one or more files by a document understanding project of the document understanding engine from a device scanning one or more corresponding paper documents into a non-standardized format, the document understanding project comprising a predefined domain set;
• automatically determining, by a generative artificial intelligence model of the document understanding engine, a domain from the predefined domain set for each of the one or more files using content of each file using a panel of large language models to vote on domain predictions, wherein the document understanding engine prompts intervention when a majority of the large language models disagree on the predictions;
• presenting, by the document understanding engine, the one or more files documents according to the domain in a user interface of the document understanding engine;
• executing a pre-annotation of the one or more files, wherein content of the one or more files is pre-annotated with interface augmentations including at least one of underlining, shapes, shading, or color-coding,
• wherein the pre-annotation utilizes a combination of a specialized model and a large language model, wherein if the specialized model does not recognize content, the large language model predicts a pre-annotation for the content.