Google Inc.

Krish Narang

Pravir Kumar Gupta

Nitin Khandelwal

Mahesh S Palekar

Arne Vansteenkiste

Ester Lopez Berga

Aakash Goel

Tayfun Elmas

Robin Martinjak

Keith Brady

Akshay Jaggi

Implementations include identifying, from a database of entries reflecting past automated assistant commands submitted within a threshold amount of time relative to a current time, particular entries that each reflect corresponding features of a corresponding user submission of a particular command. Further, those implementations include determining that the particular command is a golden command, for a particular automated assistant function, responsive to determining that: at least a threshold percentage of the user submissions of the particular command triggered the particular automated assistant function, and a quantity of the user submission of the particular command satisfies a threshold quantity. Those implementations further include, responsive to determining that the particular command is currently the golden command: processing a stream of current occurrences of submissions of the particular command to determine whether the particular automated assistant function is triggered; and when the processing indicates that a threshold quantity and/or percentage of the current occurrences fails to satisfy a potential problem threshold: causing one or more electronic alerts to be transmitted.

Low latency automated identification of automated assistant function failure

Growing community of inventors

Mountain View, CA, United States of America

Krish Narang