US 12,470,658 B2
Method for identifying contact reason of a business process discovered by a desktop analytics tool
Tal Greenberg, Bat-Hefer (IL); Simone Mimun, Givath Shmuel (IL); and Eran Roseberg, Hogla (IL)
Assigned to NICE LTD., Ra'anana (IL)
Filed by NICE LTD., Ra'anana (IL)
Filed on Feb. 1, 2024, as Appl. No. 18/429,490.
Prior Publication US 2025/0254240 A1, Aug. 7, 2025
Int. Cl. H04M 3/51 (2006.01)
CPC H04M 3/5175 (2013.01) [H04M 3/5183 (2013.01)] 9 Claims
OG exemplary drawing
 
1. A computerized-method for identifying a contact reason of a business process that has been discovered by a desktop analytics tool, said computerized-method comprising:
(i) monitoring each interaction between an agent and a customer in a contact center to:
a. generate a related interaction-transcription of the interaction and store the related interaction-transcription by an analytics Application Programming Inter (API); and
b. collect and store a sequence of user desktop actions which were operated via an application or across applications during the interaction by the agent to carry out the business process;
(ii) operating the desktop analytics tool to find a repetitive sequence of user desktop actions which were operated via the application or across the applications during the interaction; and to determine the repetitive sequence of user desktop actions as a discovered-routine for automation of the business process,
wherein each sequence of the repetitive sequence of user desktop actions is stored as an instance of the discovered-routine in a routines-datastore by the desktop analytics tool, with an associated, agent-id, start-timestamp, and end-timestamp;
(iii) retrieving a random sample of a preconfigured number of instances of the discovered-routine from the routines-datastore;
(iv) for each instance in the random sample of the preconfigured number of instances matching a related transcript-segment;
(v) identifying a contact reason of each instance by operating a first Generative (GEN) Artificial intelligence (AI) with Large Language model (LLM) with a first prompt-text including an embedded related transcript-segment,
wherein the first GEN AI with LLM has been trained to provide the contact reason based on the transcript-segment which is embedded in the first prompt-text; and
(vi) identifying the contact reason of the discovered-routine as the contact reason of the business process by operating a second GEN AI with LLM with a second prompt-text for aggregation of all the contact reasons of all instances in the random sample,
wherein the second GEN AI with LLM has been trained to provide a contact reason based on the contact reasons of all instances which are embedded in the second prompt-text.