Just-in-time inference is a term used in the fields of artificial intelligence, Industry and Factory 4.0 and automation. It describes a method in which artificial intelligence (AI) performs its calculations or predictions at the exact moment they are needed - i.e. "just in time". This saves storage space and energy because no permanent calculations and caching are required.
Imagine an AI-based system controlling the production process in a modern factory. Instead of constantly evaluating all sensor data in full, the system only analyses the most important information when, for example, a part appears on the conveyor belt. The AI calculates at lightning speed whether the part has the right shape and quality. Only then, at exactly the right time, does the just-in-time inference come into play.
This method enables efficient utilisation of resources. This is particularly advantageous on mobile devices or in factory production, as there is little computing power and memory available. Just-in-time inference thus helps companies to react faster and more efficiently to new situations - without having to upgrade expensive hardware.
Deutsch 
English (UK)