US 12,130,848 B1
Systems and methods for conversation orchestration using trained large language models
Rajkumar Koneru, Windermere, FL (US); Prasanna Kumar Arikala Gunalan, Hyderabad (IN); Thirupathi Bandam, Hyderabad (IN); Jayesh Arunkumar Jain, Bhusawal (IN); and Vishnu Vardhan Sai Lanka, Medchal Malkajgiri (IN)
Assigned to Kore.ai, Inc., Orlando, FL (US)
Filed by KORE.AI, INC., Orlando, FL (US)
Filed on Aug. 4, 2023, as Appl. No. 18/230,395.
Int. Cl. G10L 15/22 (2006.01); G06F 16/332 (2019.01); G06F 40/295 (2020.01); G06F 40/40 (2020.01); G10L 15/18 (2013.01)
CPC G06F 16/3329 (2019.01) [G06F 40/295 (2020.01); G06F 40/40 (2020.01); G10L 15/22 (2013.01); G10L 15/18 (2013.01); G10L 15/1815 (2013.01); G10L 15/1822 (2013.01); G10L 2015/221 (2013.01); G10L 2015/223 (2013.01); G10L 2015/225 (2013.01); G10L 2015/228 (2013.01)] 21 Claims
OG exemplary drawing
 
1. A method for conversation orchestration comprising:
receiving, by a conversation management framework, a plurality of conversational inputs as part of a conversation from a customer device;
generating, by the conversation management framework, a version of a data record of the conversation upon: the receiving of each of the conversational inputs; or receiving each output generated by one of a plurality of software modules communicatively coupled to the conversation management framework when the one of the software modules receives a system input from the conversation management framework, wherein the software modules comprise one or more conversational modules and one or more transactional modules;
providing, by the conversation management framework, each of the generated versions of the data record to a communication orchestrator communicatively coupled to the conversation management framework;
receiving, by the conversation management framework, for each of the generated versions of the data record, execution instructions configured as executable actions by the conversation management framework with one or more software modules from the communication orchestrator, wherein the conversation management framework is configured to communicate during the conversation with the customer device, the communication orchestrator, and the software modules comprising one or more application specific large language models and an application programming interface (API) in each of their corresponding input and output formats to provide one or more of the outputs of the software modules and wherein the communication orchestrator is a large language model trained using: a flow description comprising one or more versions of training data record; and the execution instructions correspond to the one or more versions of the training data records and comprise software code written in one or more programming languages configured to communicate with the software modules comprising the one or more conversational modules comprising the one or more application specific large language models and the one or more transactional modules comprising the API;
communicating, by the conversation management framework, with the one or more of the software modules based on the received execution instructions for each of the generated versions of the data record, wherein the communicating further comprises generating a plurality of system inputs to communicate with the one or more conversational modules and the one or more transactional modules; and
providing, by the conversation management framework, based on the communicating, one or more of the outputs of the software modules to the customer device, when the one or more of the outputs comprise one or more responses to one or more of the conversational inputs, wherein at least two or more of the generated versions of the data record comprise: one or more of the conversational inputs, one or more of the execution instructions, and one or more of the outputs.