Generative artificial intelligence based approach for processing requests for information about electronic documents

A system for automatically generating a response to a current request for information about an electronic document associated with a particular domain of knowledge and prepared using a software application, the system comprising:

• a memory including computer executable instructions;
• and
• a processing configured to execute the computer executable instructions and cause the system to:
• interpret, using a first language processing machine learning model, the electronic document to obtain interpreted data about the electronic document;
• select, based on the current request, a subset of feedback data included in a first knowledge store configured to store feedback data for prior responses to prior requests for information about the electronic document or a different electronic document;
• select, based on the current request, a subset of domain data stored in a second knowledge store configured to store domain data associated with the particular domain of knowledge;
• inject a prompt into the first language processing machine learning model or a second language processing machine learning model, the prompt based on the current request, the subset of feedback data, the subset of domain data, and the interpreted data;
• and
• generate, via the first language processing machine learning model or the second language processing machine learning model, the response to the current request based on the injecting.

A non-transitory computer-readable medium comprising instructions to be executed in a processing system to automatically generate a response to a current request for information about an electronic document associated with a particular domain of knowledge and prepared using a software application, wherein the instructions, when executed, cause the processing system to perform a method comprising:

• interpreting, using a first language processing machine learning model, the electronic document to obtain interpreted data about the electronic document;
• selecting, based on the current request, a subset of feedback data included in a first knowledge store configured to store feedback data for prior responses to prior requests for information about the electronic document or a different electronic document;
• selecting, based on the current request, a subset of domain data stored in a second knowledge store configured to store domain data associated with the particular domain of knowledge;
• injecting a prompt into the first language processing machine learning model or a second language processing machine learning model, the prompt based on the current request, the subset of feedback data, the subset of domain data, and the interpreted data;
• and
• generating, via the first language processing machine learning model or the second language processing machine learning model, the response to the current request based on the injecting.