T-mobile Usa, Inc.

Robert Stamm

James Ellison

Joel Werdell

Phi Nguyen

Kevin Lau

Timothy Shelton

Rudy Bourcelot

Ryan Yokel

Disclosed here is a system to automatically predict and resolve issues within a telecommunication network. Initially, the system builds a service registry to store dependence information within the network, which can include software components and hardware components. Various components of the network create logs of their operations. Machine learning models examine the logs and detect any issues. Upon detecting an issue or abnormal event, the system can automatically resolve the issue by determining the most similar issue occurring previously and determining a solution that resolved the previous most similar issue. In addition, the system can propagate the fix to dependent systems and/or notify the dependent systems of the issue.

Predicting and resolving issues within a telecommunication network

This disclosure describes techniques that facilitate selectively interacting with a computing resource based on receipt of an incoming voice command. Particularly, a voice control integration system may parse content of an incoming voice command to authenticate an identity of the client, and further determine an intended meaning of the incoming voice command. In doing so, the voice control integration system may interact with a computing resource to perform an action that fulfills a client request. Computing resources may be associated with service providers or client devices of a client. Further, the voice control integration system may authenticate a client identity based on a one or two-factor authentication protocol, of which one may correspond to a biometric analysis of the incoming voice command. Further, a second-factor of the two-factor authentication protocol may be implemented via a voice interaction device, another client device accessible to the client, or a combination of both.

Voice control and telecommunications service integration

This disclosure describes techniques that facilitate selectively automating a response to a client request for vendor services based on an intended meaning of the client request, a category of vendor services associated with the client request, and one or more factors such as an urgency index and a health index. The vendor services system may parse audio, text, and image data of a client request to determine an intended meaning. In some examples, the vendor services system may automate a response that resolves the client request, or automate a response that requests additional information from the client device. Alternatively, or additionally, the vendor services system may direct the client request to a vendor representative for further processing. In doing so, the vendor services system may monitor dialogue between the client and vendor representative, and selectively provide the vendor representative with support data that may help resolve the client request.

Real-time integration of machine intelligence into client messaging platforms

A text representation is received from a virtual assistant application on a user device. The text representation may be generated via a speech-to-text engine of the virtual assistant application from audio speech spoken by a customer. Device diagnostic data of the user device is also received from the virtual assistant application. An identifier of the customer is placed in an assistance queue. At least information in the text representation and the device diagnostic data is analyzed to determine an issue associated with the user device and a solution for resolving the issue, so that the solution is applied. In response to the identifier of the customer reaching a front of the assistance queue, session state information that includes at least the text representation and a description of the issue is provided to a support application. A voice support session is initiated between the support application and the virtual assistant application.

Device-based audio processing for enhanced customer support

Techniques for identifying a likely subject of a user contact with a user support network and routing the user contact to an agent or an application that is determined to have relevant experience with the subject or issue the user is experiencing. To identify a likely subject of the user contact, data is gathered from a user device as well as from secondary user devices located near the user device. Also, data related to network resources the user device is using, history data of a user's previous contacts with the user support network, and account data associated with the user device is also collected. As a result, the user contact is routed to a knowledgeable agent who is likely to resolve the user's issue without further routing or consultation, thus saving network bandwidth, user device resources, agent time, user time, etc.

Predictive service for smart routing

Cellular networks regularly operate with trusted devices, which typically are trusted because of the integration of a SIM card therewith. Untrusted devices typically do not interface with a SIM card, and rely on user input through traditional user interfaces for authentication. Recently, the use of hands-free, always-on digital assistant devices have become more common. Such devices typically have only a voice user interface that may be used by a number of people in close proximity to the device. Particular problems arise in such a scenario when a user wants to access a secure service that requires user authentication. Such problems are addressed with multiple techniques described herein.

Growing community of inventors

Sammamish, WA, United States of America

Robert Stamm