Record, reflect, then repeat using async video. The training experience covers various modules (Eye Contact, Voice Tone, etc) and encourages users to mess up as much as necessary.