Abstract: A system and method provide a trained model that uses vectorized word embeddings that are averaged or summed to form representations for sentences and phrases. The representations are processed in a Siamese neural network including multiple LSTM stages to find semantically related matches in catalogs for non-catalog queries. The model is trained using catalog data and randomized data using a contrastive loss function to generate similarity metrics for catalog-non-catalog pairs.

Abstract: An automation optimization engine is disclosed that provides a closed loop nodal analysis on a process performance tree. By analyzing an observed process and identifying automatable tasks from non-automatable tasks from the observed process, the automation optimization engine then applies machine learning techniques to generate a recommendation report identifying steps for implementing in the observed process to convert the non-automatable tasks into automatable tasks. Optimization is thus achieved to close the loop to further automate the observed process.

Abstract: An automation optimization engine is disclosed that provides a closed loop nodal analysis on a process performance tree. By analyzing an observed process and identifying automatable tasks from non-automatable tasks from the observed process, the automation optimization engine then applies machine learning techniques to generate a recommendation report identifying steps for implementing in the observed process to convert the non-automatable tasks into automatable tasks. Optimization is thus achieved to close the loop to further automate the observed process.

Abstract: A robotic operations control system enables seamless distribution of tasks among a blended workforce that includes human employees and robot assistants. The robot assistants include an intelligent billing assistant (IBA), an intelligent collections assistant (iNCA), a dispute resolution assistant (DRA) and a cash processor. Each of the assistants includes a workflow manager that receives work items to be processed by the robot assistant based on the priority levels assigned to the work items. The IBA includes a dispute prediction model that predicts the likelihood of dispute for a work item and automatically generates a customized checklist. The iNCA includes an AI prioritizing assistant that identifies priorities for work items to be processed. The DRA includes an AI assisted investigator that expedites investigations of disputes by proactively gathering information for a human employee to review and mark the dispute as valid, invalid or for query.

Abstract: An automation identification diagnostic tool (AIDT) is disclosed that records and analyzes actions within a program that are taken by an agent to accomplish certain tasks. By recording the action steps to accomplish a task in a recording session data file, the AIDT is then able to analyze the recording session data file to identify steps for conversion to automated tasks.

Abstract: A robotic operations control system enables seamless distribution of tasks among a blended workforce that includes human employees and robot assistants. The robot assistants include an intelligent billing assistant (IBA), an intelligent collections assistant (iNCA), a dispute resolution assistant (DRA) and a cash processor. Each of the assistants includes a workflow manager that receives work items to be processed by the robot assistant based on the priority levels assigned to the work items. The IBA includes a dispute prediction model that predicts the likelihood of dispute for a work item and automatically generates a customized checklist. The iNCA includes an AI prioritizing assistant that identifies priorities for work items to be processed. The DRA includes an AI assisted investigator that expedites investigations of disputes by proactively gathering information for a human employee to review and mark the dispute as valid, invalid or for query.

Abstract: An automation optimization engine is disclosed that provides a closed loop nodal analysis on a process performance tree. By analyzing an observed process and identifying automatable tasks from non-automatable tasks from the observed process, the automation optimization engine then applies machine learning techniques to generate a recommendation report identifying steps for implementing in the observed process to convert the non-automatable tasks into automatable tasks. Optimization is thus achieved to close the loop to further automate the observed process.

Abstract: Implementations for consignment booking include actions of providing a user interface including a plurality of consignment data entry fields that include a pickup location entry field configured to receive data identifying a pickup location for a consignment, a delivery destination entry field configured to receive data identifying a delivery destination for the consignment, and a content entry field configured to receive data identifying a content of the consignment to be delivered, retrieving the data identifying the pickup location, the delivery destination and the content, transmitting a call signal to be transmitted to a mobile computing device of each agent in a plurality of agents, determining a respective location of each agent in the plurality of agents based on a response to each call signal, and selecting an agent from the plurality of agents based at least in part on the respective agent locations with respect to the pickup location.

Abstract: An automation identification diagnostic tool (AIDT) is disclosed that records and analyzes actions within a program that are taken by an agent to accomplish certain tasks. By recording the action steps to accomplish a task in a recording session data file, the AIDT is then able to analyze the recording session data file to identify steps for conversion to automated tasks.