In plain English
Training is the process of using data to adjust a model. The trained model can then make predictions or produce outputs from new inputs, a process called inference. Teams evaluate it on separate data to check how well its learned patterns carry over.
Example
A startup trains a model on past support messages labeled as billing or technical questions. When a new message arrives, the trained model predicts its category to help route it to the right team.