By avoiding the operator validation we gained 10% of the assembly time, and divided the training time by 2