Amazon Technologies, Inc.

Juliana Saussy

Pasquale DeMaio

Anurag Windlass Gupta

Mehul A Shah

Mehul Y Shah

Prajakta Datta Damle

Stephen Michael Ash

Timothy Jones

Andrew Borthwick

Abhishek Sharma

Shehzad Qureshi

Pedrito Uriah Maynard-Zhang

Alaykumar Navinchandra Desai

Sergei Dobroshinsky

Vamsi Polapragada

Sonu Kumar Mishra

Chethan Kommaranahalli Rudramuni

Adam Lawrence Joseph Heinermann

Joseph Patrick Sullivan

Jon R Jay

Joseph Daniel Sullivan

Thomas Boyd Johnston

Kevan Mah

Naji Shaheimi Shafi

Akshay Isaac Lazarus

Preetham Gowda

Systems and methods are described relating to distributing agents to different queues provided by a call or contact center. A contact service may define a number of different queues for processing different customer requests, such as may be routed to different agents, with each queue associated with a priority value (higher priority get assigned agents first) and a visibility threshold (how long a request will sit before it is picked up by an agent). The described techniques add a time-to-service level (SL) goal, per queue to determine when queues are over and underperforming. In the case queues are overperforming, various techniques, including control systems approaches may be used to determine adjustments to the visibility threshold (e.g., at least one of increasing the thresholds for overperforming queues and decreasing the thresholds for underperforming queues) to increase performance in meeting the customer SL goals.

Service level based routing using visibility thresholds

Specified performance attributes may be used to configure machine learning transformations for ETL jobs. Performance attributes for a machine learning pipeline that applies a model to as part of a transformation for an ETL job may be used to configure a parameter in a stage of the machine learning pipeline. The configured stage may then be used when training the model. The trained machine learning pipeline may then be applied as part of a transformation operation included in an ETL job performed by the ETL system.

Using specified performance attributes to configure machine learning pipepline stages for an ETL job

An adaptive virtual assistant system can be configured to change an attribute of a virtual assistant based on user responses, environmental conditions, and/or topics of discussion. For example, the virtual assistant system can determine, based at least in part on user data, a communication profile that is associated with the virtual assistant and determine first communication data comprising a first communication attribute based on the communication profile. In some instances, the system can transmit the first communication data to a user device and receive, from the user device, input audio data representing a user utterance. Based at least in part on the input audio data, the system can determine second communication data comprising a second communication attribute and transmit the second communication data to the user device.

Adaptive virtual assistant attributes

Technology is described for routing communications between contacts and agents. Associations may be generated between agents and contacts who intend to communicate with the agents. Weightings for the associations between the agents and the contacts may be determined using a machine learning model to produce weighted associations. The weightings may represent a predicted interaction metric between the agents and the contacts. Selected pairs of agents and contacts may be determined by applying matching rules to the weightings. Communications may be routed between the contacts and the agents in accordance with the selected pairs of agents and contacts.

Communication routing between agents and contacts

This disclosure describes management of a contact center executing in a service provider network. The management may be performed by using one or more trained deep learning/machine learning (ML) models. The ML models may be trained using metrics and data gathered from the various services and systems in the service provider network used to implement the contact center. The trained models may be used for forecasting staffing needs, e.g., agents, for the contact center, scheduling agents, detecting anomalies with regard to customer traffic, e.g., received communications from customers, at the contact center, adherence of agents to the scheduling, and dynamically routing of received communications to appropriate agents (e.g., based on skills of the agent and other factors) for handling of the communications.

Dynamic communication routing at contact centers

Using specified performance attributes to configure machine learning pipeline stages for an ETL job

Systems for providing a contact center system detect patterns in the behavior of service agent users and customers having a customer service issue. The system records activity data describing interactions of the users with a virtualized contact center and with each other; the activity data is aggregated and then analyzed (e.g., using a machine learning engine) to measure one or more metrics that define a behavior pattern. The analysis identifies an intent expressed by customers having a common issue, and further identifies attributes of the intent as well as the actions taken by the service agents in response to the intent. The system uses the analysis to generate a workflow of potential interactions that resolve the issue. The system automates the workflow by generating and deploying a computerized service agent configured to interact with customers of a contact center and execute the workflow.

Automation of contact workflow and automated service agents in contact center system

Growing community of inventors

Seattle, WA, United States of America

Juliana Saussy