US 12,147,788 B2
System and method for intelligently generating code for use in integrating automated response generating systems with non-API applications
Donna Maria Welch, Bluffton, SC (US); Sudhakar Balu, Chennai (IN); Srinivasa Dhanwada, Hyderabad (IN); and Siva Kumar Paini, Hyderabad (IN)
Assigned to Bank of America Corporation, Charlotte, NC (US)
Filed by BANK OF AMERICA CORPORATION, Charlotte, NC (US)
Filed on May 11, 2022, as Appl. No. 17/741,953.
Prior Publication US 2023/0367557 A1, Nov. 16, 2023
Int. Cl. G06F 8/34 (2018.01); G06F 9/451 (2018.01); G06F 16/954 (2019.01)
CPC G06F 8/34 (2013.01) [G06F 9/451 (2018.02); G06F 16/954 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A system comprising:
a memory;
a display; and
a hardware processor communicatively coupled to the memory and the display, the hardware processor configured to:
determine that the display is displaying an element associated with an application, the element comprising at least one of a graphical user interface and a web browser;
in response to determining that the display is displaying the element:
record a video of the display, the video depicting a process performed using the element, to obtain a first set of information from the application;
extract, from the video of the element, metadata associated with the process depicted in the video, wherein extracting metadata is performed for an action identified in the video, the action including at least one of: logging into an application, navigating to a page displayed by the application, or entering text into a field displayed on the page;
using a trained machine learning model, generate, based at least in part on the extracted metadata, a set of instructions for interfacing with the application, the set of instruction comprising at least one of a custom API or a set of headless browser steps, wherein, when executed by the hardware processor, the set of instructions are configured to cause the processor to interface with the application to obtain the first set of information from the application, wherein the machine learning model is trained based on previously extracted metadata and associated set of instructions; and
store the set of instructions in the memory.