depends on #6 to complete before proceeding - reduces delays from latency when making multiple inferences - applies to web interface only
