Personalized help using user behavior and information

A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors, cause the one or more processors to perform a method for improving a user experience of an application, the method comprising:

• receiving user activity information related to activities of a user during use of the application;
• providing the user activity information as input to a predictive model, wherein the predictive model has been trained through a supervised machine learning process using features derived from historical activity data associated with labels derived from historical solution data associated with a plurality of users to determine actionable solutions to issues related to use of the application, wherein the historical solution data comprises data related to actions that historically resulted in resolutions of issues indicated by the historical activity data, and wherein the historical solution data indicates that modifying preciously-provided login information on a particular user interface screen historically resolved a particular issue related to automatically logging in, by the application, to a user account associated with a separate application;
• determining, based on output from the predictive model, an actionable solution to the particular issue, wherein the particular issue is related to use of the application by the user;
• providing the actionable solution to the user via a user interface associated with the application, wherein the actionable solution is provided without being initiated by the user, and wherein providing the actionable solution to the user via the user interface associated with the application comprises:
• replacing a current user interface screen displayed within the user interface with the particular user interface screen comprising login information that the user previously provided via the particular user interface screen, wherein the login information corresponds to an account of the user associated with a different application that is separate from the application;
• and
• prompting the user to modify the login information that the user previously provided via the particular user interface screen;
• and
• determining whether the particular issue related to use of the application by the user was resolved by the actionable solution based on clickstream data or log data related to performing the actionable solution within the application, wherein the predictive model is re-trained based on the determining whether the particular issue related to use of the application by the user was resolved by the actionable solution.

A system, comprising one or more processors and a non-transitory computer-readable medium comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform a method for improving a user experience of an application, the method comprising:

• receiving user activity information related to activities of a user during use of the application;
• providing the user activity information as input to a predictive model, wherein the predictive model has been trained through a supervised machine learning process using features derived from historical activity data associated with labels derived from historical solution data associated with a plurality of users to determine actionable solutions to issues related to use of the application, wherein the historical solution data comprises data related to actions that historically resulted in resolutions of issues indicated by the historical activity data, and wherein the historical solution data indicates that modifying previously-provided login information on a particular user interface screen historically resolved a particular issue related to automatically logging in, by the application, to a user account associated with a separate application;
• determining, based on output from the predictive model, an actionable solution to the particular issue, wherein the particular issue is related to use of the application by the user;
• providing the actionable solution to the user via a user interface associated with the application, wherein the actionable solution is provided without being initiated by the user, and wherein providing the actionable solution to the user via the user interface associated with the application comprises:
• replacing a current user interface screen displayed within the user interface with the particular user interface screen comprising login information that the user previously provided via the particular user interface screen, wherein the login information corresponds to an account of the user associated with a different application that is separate from the application;
• and
• prompting the user to modify the login information that the user previously provided via the particular user interface screen;
• and
• determining whether the particular issue related to use of the application by the user was resolved by the actionable solution based on clickstream data or log data related to performing the actionable solution within the application, wherein the predictive model is re-trained based on the determining whether the particular issue related to use of the application by the user was resolved by the actionable solution.