A streaming server which streams an application to a client computer (“endpoint”), as well as the client on which the streamed application runs, makes predictions as to what sections of the application the client is likely to execute in the future. Upon receipt of an indication (e.g., from a system administrator) of a planned service outage of the server or the network, the server transmits the application content that is predicted to be needed by the client during the outage in order to continue executing the application without interruption. The client receives and caches the content. Provided that the prediction is sufficiently accurate, the client can continue to seamlessly execute the application during the service outage.